Skip to main content

Full text of "Energetics of tropical hummingbirds : effects of foraging mode, competition, and resource availability"

See other formats


ENERGETICS OF TROPICAL HUMMINGBIRDS: 

EFFECTS OF FORAGING MODE, COMPETITION, 

AND RESOURCE AVAILABILITY 



HARRY MORGAN TIEBOUT III 



A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL 

OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT 

OF THE REQUIREMENTS FOR THE DEGREE OF 

DOCTOR OF PHILOSOPHY 

UNIVERSITY OF FLORIDA 



MAY 1990 



Copyright 1990 

by 

Harry Morgan Tiebout III 



ACKNOWLEDGMENTS 



This project would not have been possible without the 
help of many individuals and institutions. I thank first 
Kristin Brugger for everything, including help with 
experimental design, mist-netting, and editorial review. 

To Peter Feinsinger I owe an incalculable debt. Pete's 
own doctoral research got me interested in tropical 
hummingbirds while I was an undergraduate, and he has been an 
unwavering source of support, encouragem.ent , and enthusiasm 
throughout my graduate career. The other persons who served 
on my committee have provided invaluable assistance as well, 
from helping me ro frame my initial proposals to reviewing 
the final product. My deepest appreciation goes to Martha 
Crump, John Eisenberg, Jack Ewel, Harvey Lillywhite, and Jack 
Putz. Many other faculty and graduate students in my 
department helped at various stages, including John Anderson, 
William Busby, Jack Kaufmann, Sharon Kinsman, Alan Masters, 
Brian McNab, Greg Murray, Ken Prestwich, Gene Spears, and 
Kathy Winnett-Murray . I also thank Dave Evans, Bill Karasov, 
and Ken Nagy who helped me to learn the techniques for using 
isotopic markers. 



Ill 



I am grateful to the many people who made my research in 
Costa Rica possible. Jose Arturo Leon and the rest of the 
OTS Costa Rican Office handled the worst of my logistical 
nightmares. In the field I was ably assisted by Lisa Ellis, 
Tim Kertt, Dwight Lawson, Cindy Lordan, and Doug Mason. 
Cathy Langtimm guarded my sanity. 

Finally, I offer my sincerest thanks to the residents of 
Monteverde, Costa Rica, who allowed me to work and live in 
their wonderful community. I owe a special gratitude to John 
and Doris Campbell, Michael and Patricia Fogden, Adrian and 
Turid Forsyth, Tomas and Lindi Guindon, Benito and Tonio 
Guindon, Wolf and Lucky Guindon, Bill and Barb Haber, Meg and 
Rich LaVal, Bob Law, Dave McDonald, Susie Newswanger, Alan 
Pounds, the Eston Rockwell family, the Paul Smith family, the 
Joe and Jean Stuckey family, Stella Wallace, and Willow 
Zuchowski . 

My research was supported by the Department of Zoology 
and the Division of Sponsored Research at the University of 
Florida, Mark William Hoppe, NSF grants to P. Feinsinger 
(BSR-8605043) and P. Feinsinger and H. Tiebout (BSR-870 1012 ) , 
The Jesse Noyes Foundation (administered through The 
Organization for Tropical Studies), Sigma Xi, Elizabeth M. 
Tillson, and The Tinker Foundation (administered through The 
Center for Latin Ajnerican Studies) . 



IV 



TABLE OF CONTENTS 



page 

ACKNOWLEDGMENTS ' iii 

ABSTRACT viii 

CHAPTERS 

1 INTRODUCTION 1 

Overview 2 

Neotropical Nectarivore Guilds 2 

Mechanisms of Guild Structure 3 

Determinants of Energetic Costs and 

Benefits in Hummingbirds 5 

Energy Regulation and Foraging Decisions 6 

Energy Management: Time Budgets and 

Territory Size 7 

General Approach and Goals of this Study 8 

Foraging Mode 9 

Physiological Constraints on Rates 

of Food Processing 13 

Competition for Food 14 

Food Availability and Dispersion 17 

Food Dispersion and Competition 18 

2 GENERAL MATERIALS AND METHODS 21 

Hummingbird Capture and Care 21 

Energy Budgets 22 

Behavior 2 9 

Analyses and Statistics 31 

Data Interpretation and Presentation 33 



V 



FOOD PASSAGE RATES IN HUMMINGBIRDS 3 6 

Introduction 35 

Methods 39 

Analysis 40 

Results 42 

Discussion 44 

Tests of Predictions 1 and 2: Patterns of 

Excretion Rates 44 

Evaluation of Predictions 3 and 4 : Constraints on 

Intake and Foraging Time 45 

Application to Experimental Design 47 

Conclusions 49 

INTER- AND INTRASPECIFIC COMPETITION FOR A 
DEFENSIBLE AD LIBITUM FOOD SOURCE 54 

Introduction 54 

Overview 54 

Goals and Predictions 55 

Methods 59 

Experimental Cage and Protocol 59 

Dependent Variables 50 

Analysis 53 

Results 64 

Solitary Birds 64 

Paired Birds 65 

Discussion 73 

THE EFFECTS OF FOOD AVAILABILITY AND 

INTERFLOWER DISTANCE UNDER CONDITIONS OF 

NO COMPETITION 98 

Introduction 93 

Background and Goals 98 

Energy Management 99 

Experimental Design and Predictions 100 

Materials and Methods 102 

Flight Cage 102 

Protocol 103 

Dependent Variables and Analysis 105 

Results 107 

Behavioral Responses IO7 

Energetic Responses 108 

Discussion 11]_ 

General Responses to Foraging Constraint Ill 

Interspecific Differences 117 



vx 



6 VALIDATION OF THE DOUBLY LABELED WATER .METHOD 

(3hh18o) for measuring WATER FLUX AND CO2 

PRODUCTION IN AMAZILIA SAUCEROTTEI 125 

Introduction 125 

Materials and Methods 128 

Protocol 128 

Calculations 129 

Results 133 

Total Body Water 133 

DLW versus BAL Estimates 134 

Discussion 135 

7 INTERSPECIFIC COMPETITION FOR LIMITED FOOD: 

EFFECTS OF FLOWER DISPERSION 140 

Introduction 140 

Materials and Methods 143 

Experimental Flight Cage 143 

Protocol 144 

Experimental Design and Analyses 145 

Results 14 9 

Behavior 149 

Energetics 153 

Discussion 155 

Behavioral and Energetic Responses 155 

Relative Energetic Success of Each Foraging Mode... 158 



8 GENERAL DISCUSSION AND CONCLUSIONS 173 

Implications of Results for Amazilia and 

Chlorostilhon in the Wild 173 

Physiological Constraints to Feeding 176 

Intra- and Interspecific Competition 177 

Food Availability and Interf lower Distance 181 

Food Dispersion and Interspecific Competition 186 

General Trends Across Experiments . 191 

The Mechanistic Approach to Understanding 

Natural Guilds 195 

LITERATURE CITED 201 

BIOGRAPHICAL SKETCH 216 



vii 



W-^V-^^TMil'l - — 



Abstract of Dissertation Presented to the Graduate School 

of the University of Florida in Partial Fulfillment of the 

Requirements for the Degree of Doctor of Philosophy 



ENERGETICS OF TROPICAL HUMMINGBIRDS: 

EFFECTS OF FORAGING MODE, COMPETITION, 

AND RESOURCE AVAILABILITY 

By 

Harry Morgan Tiebout III 

May 1990 



Chairman: Peter Feinsinger 
Major Department: Zoology 



In the neotropics, short-billed humm.ingbirds with 
contrasting foraging modes presumably differ in their 
energetic costs and benefits while foraging for their primary 
energy source, floral nectar. Relative differences in 
energetic success should depend upon nectar availability, 
including reward per flower, flower dispersion, and 
competitor pressure. I tested this hypothesis by comparing 
the behavior and energetics of a territorialist , Amazilia 
saucerotteif and a low-reward trapliner, Chlorostilbon 
canivetii, under prescribed conditions of food availability. 
In a series of cage experiments, I measured energy budgets 
(intake = expenditure + storage) as mass-specific rates, 
using gravimetric or doubly labeled water (DLW) methods . I 



vixi 



assumed that energetic success was proportional to rates of 
energy storage. 

First, I examined the consequences of sharing a feeder 
(artificial flower) containing an unlimited amount of food 
(sucrose solution) . Birds tested in both conspecific and 
heterospecif ic pairs generally experienced increased energy 
expenditure and decreased energy storage compared to solitary 
birds. Because the feeder was easily defensible, the 
behaviorally dominant Amazllia were more successful 
energetically in heterospecif ic pairs than in conspecific 
pairs, whereas the reverse was true for Chlorostilbon, 
subordinate to Amazilia . I then tested solitary birds at two 
food levels (high-reward, low-reward) and two perch-to-feeder 
distances (near, far) . When feeders were low-reward, both 
species were energetically unsuccessful at either distance, 
with Chlorostilbon experiencing lower energy storage rates 
than Amazllia . Using the DLW method, I tested the effects of 
feeder dispersion on heterospecif ic pairs competing for 
limited food. When feeders were clumped (hence defensible), 
aggressive behavior by Amazllia restricted food intake by 
Chlorostilbon, resulting in higher energetic success for the 
territorialist . When feeders were dispersed (hence 
indefensible) , the trapliner had higher overall intake rates 
and per visit intake rates than the territorialist . 
Nevertheless, Chlorostilbon again experienced lower energetic 
success than Amazllia . 



IX 



In general, Chlorostilbon had higher energy expenditure 
rates than Amazllia, due to higher resting metabolic rates 
and greater flight time. Thus, whenever intake was limited, 
due to low-reward feeders or to interference by Amazilla, the 
trapliner experienced lower energetic success than the 
territorialist . Across all experiments, Amazllia and 
Chlorostilbon appeared energetically specialized for 
different conditions of nectar availability and competition. 



X 



CHAPTER 1 
INTRODUCTION 



Guilds of neotropical nectar-feeding birds often contain 
many species and display rapid structural changes correlated 
with shifts in climate or food resources. Nevertheless, the 
actual mechanisms responsible for temporal changes in species 
composition are poorly known. Previous research on 
hummingbird assemblages (a) shows that populations track 
their food resources over time and space and (b) suggests 
that species with distinct foraging modes experience 
different energetic consequences from particular dispersions 
or densities of flower nectar. This establishes the central 
proposition of my study: differential energetic 
consequences, due to species-specific differences in foraging 
mode, partly determine each species' presence at a particular 
site . 

The goal of my dissertation was to evaluate part of this 
proposition by determining whether co-occurring, short-billed 
hummingbird species representing two radically different 
foraging modes differ in their energetic costs and gains 
under a range of foraging conditions. Such differential 
energetic response is an important and necessary precondition 
for confirming the proposition, though it is not in itself 



sufficient. I focused on several variables thought to affect 
foraging costs and gains. Through a series of 5 experiments, 
I investigated (a) the flexibility of behavioral and 
physiological responses of individuals to each variable, and 
(b) species-specific responses that relate to their different 
foraging modes. I conclude this dissertation with a 
synthesis of experimental results, discussing their relevance 
to the energetics of wild hummingbirds and possibly to the 
structure of nectarivore guilds . 

Overview 

Neotro pical Nectarivore Guilds 

Many field investigations describe compositions of 
guilds {sensu Root 1967) of both nectarivorous birds and 
their food plants at neotropical sites (e.g.. Snow and Snow 
1972,1980, Colwell 1973, Feinsinger 1974, 1976, 1978, 1980, 
Stiles 1975, 1980, 1981, 1985, Wolf et al . 1976, Des Granges 
1979, Kodric-Brown et al . 1984, Feinsinger et al. 1985, 1987, 
1988) . In general, avian assemblages share several 
attributes. First, flower-visiting bird species exhibit 
several distinct foraging modes that correlate with 
characteristic morphologies. These modes, also called roles 
or foraging strategies, may be differentially efficient at 
utilizing various food resource states (Feinsinger and 
Chaplin 1975, Wolf et al. 1976, Feinsinger and Colwell 1978, 



Feinsinger et al. 1979, Brown and Bowers 1985). Second, 
species in the guild are typically restricted to the family 
Trochilidae (Stiles 1981. Hence, differences in foraging 
modes are likely not of phylogenetic origin. Third, 
populations occupying successional habitats often may be near 
carrying capacity and thus facing energy limitation 
(Montgomerie 1979, Montgomerie and Gass 1981) . Finally, 
guild structure can be extremely dynamic at a given site. 
Species composition may change over seasons or even days, and 
such changes are often correlated with fluctuations in the 
food base (Colwell 1973, Feinsinger 1976, 1980, Gass and 
Lertzman 1980, Stiles 1980, Murcia 1987, Feinsinger et al. 
1985) . 

Mechanisms of Guild Structure; 

Guild composition is defined by the population levels of 
the member species, which are determined by rates of birth, 
death, em.igration, and immigration. Each of these rates can 
respond to fluctuations in the food base, yet the actual 
mechanism (s) by which member species are differentially 
affected has not been docum.ented. Because member species 
share a com.mon food base that may be limited, the energetic 
costs and benefits associated with food acquisition probably 
vary with fluctuations in the food base. If hummingbird 
species with different foraging modes differ in their 
efficiencies of food acquisition, then at a given resource 



state (e.g., a clumped distribution of high-reward flowers) 
certain guild meiiLbers may be favored energetically, meaning 
they meet their energy requirements more easily than others. 
When an animal meets its energy requirements (defined 
operationally in Chapter 2) it experiences "energetic 
success;" when it does not meet these requirements, it 
experiences "energy stress" or is "energetically 
unsuccessful." As the array of available resource states 
changes over time (e.g., from a clumped to a widely dispersed 
flower distribution) , different guild members may in turn be 
favored energetically. 

Birds that are energetically successful are more likely 
to remain in an area (Gass and Lertzman 1980) than are birds 
that are energy stressed. Reproduction can be similarly 
affected, with those species enjoying higher energetic 
success also producing relatively more offspring. Thus, the 
changing structure of the guild may reflect the changing 
energetic success of each of its member species. 

Factors other than energetic success alone can determine 
local species composition. For example, a species' presence 
may reflect the availability of nesting sites or low predator 
density (Miller and Gass 1985). Regardless of the other 
factors attracting hummingbirds to an area, birds will not 
remain for long if they cannot meet their energy needs . 
Thus, although energetic success may not be sufficient to 
explain the presence of a given species, it is a necessary 
precondition . 



Determgnants of Energetic Costs and Benefits in Hnminingbi rds 

Many studies, conducted primarily in the laboratory, 
document numierous factors that influence the energetics of 
hummingbirds. Metabolic expenditures of foraging hummingbirds 
can depend on body size (MacMillen and Carpenter 1977), wing 
disc loading (Epting and Casey 1973, Greenewalt 1975), power 
for hovering flight (Epting 1975, 1980, Hainsworth and Wolf 
1972c), altitude (Berger 1974b, 1978, Feinsinger et al . 
1979), interflower distance (Wolf et al. 1975), and ambient 
temperature (Beuchat et al . 1979, Schuchmann 1979, Schuchmann 
and Schmidt-Marloh 1979a) . Efficiency at extracting nectar 
may vary with bill and tongue length (Ewald and Williams 
1982, Kingsolver and Daniel 1983) , length and shape of flower 
corolla (Hainsworth 1973, Wolf et al . 1972, 1976), nectar 
volume and concentration (Hainsworth and Wolf 1972a, 1976, 
Montgomerie 1984), floral array (Hainsworth et al . 1983), and 
possibly physiological constraints on food processing rates 
(Diamond et al . 1986, Karasov et al . 1986). Inter- and 
intraspecif ic competition among birds may also limit food 
intake and affect rates of energy expenditure (discussed 
below) . Together these studies indicate that energetic costs 
and benefits vary both with physiological and morphological 
characteristics of hummingbirds and with the resource states 
of their nectar food base. Because hummingbird species can 



vary markedly in these characteristics, species should differ 
in their energetic responses to changes in the food base. 

It is not surprising that so many environmental factors 
influence the energetics of hummingbirds. As a group, the 
Trochilidae have among the smallest body sizes and the 
highest mass-specific metabolic rates of any endotherm, with 
the possible exception of the Soricidae,' class Mammalia 
(Johnsgard 1983) . Hummingbirds have one of the most 
energetically expensive foraging modes (hovering flight), 
relatively low thermal inertia, high surface-to-volume ratio, 
and high body temperature. Thus, in order to fuel high 
activity and maintenance costs, they m>ust maintain high rates 
of energy intake relative to body mass. In addition, due to 
their extremely low energy storage capacity (see Paladino 
1989), to survive hummingbirds must be able to regulate their 
energy budgets precisely even over relatively short periods 
of time (Calder 1974) . 

Energy Regulation and F oraging Decisions 

Recent investigations demonstrate the importance of 
energetics to foraging decisions made by hummingbirds. 
Studies of optimial foraging suggest that hummingbirds follow 
certain "rules" about which patches to visit and how to 
forage within chosen patches (Pyke 1978, 1981a, 1981c, Gass 
and Montgomerie 1981) . Evidence from nectar assimilation 
rates (Hainsworth 1974), crop volume (Hainsworth and Wolf 



1972a), and the costs of transporting a meal. once it has been 
consumed (DeBenedictis et al . 1978) allow estimates of 
optimal meal size. When hummingbirds are allowed to choose 
among artificial flowers that vary in energetic cost and 
benefit parameters, birds appear to select those flowers that 
enable them to maximize net energy gain (Hainsworth 1978, 
1981b, Hainsworth et al . 1981, Montgomerie et al . 1984). 
Such choice experiments also suggest several possible 
physiological mechanisms of energy regulation. The above 
studies establish that energetic costs and benefits are very 
important to hummingbirds . Hummingbirds appear to monitor 
short-term energy intake, respond to current energy 
expenditure, and make decisions to maximize energy intake 
(gross or net) or foraging efficiency (energy intake per 
energy expended in foraging) (Gass and Montgomerie 1981) . 

Energy Management: Time Budgets and Territory Size 

Field studies on territorial species indicate that 
hummingbirds quickly adjust energy intake relative to 
expenditure. Food availability and intruder density can 
influence territory size and foraging success (Ewald and 
Carpenter 1978, Copenhaver and Ewald 1980, Carpenter et al. 
1983, Hixon et al. 1983, Paton and Carpenter 1984). Time 
budgets from territorial birds (Wolf and Hainsworth 1971, 
Carpenter et al . 1981) provide estimates of energy budgets 
(e.g., Pearson 1954, Stiles 1971, Wolf and Hainsworth 1971, 



Hainswcrth and Wolf 1972b) , and indicate that hummingbirds 
are able to adjust behavior rapidly when competitor pressure 
and food availability change. Extrapolation from time 
budgets suggests that their energy budgets may be influenced 
as well. 

Throughout my study I use the term "energy management" 
to denote the behavioral and physiological responses of an 
individual that determine its total energy budget (intake, 
expenditure, and storage; as defined in Chapter 2) . This 
term avoids several problems inherent in common expressions 
such as "maintaining energy balance" and "balancing" energy 
budgets. First, these expressions imply that there exists a 
single correct budget that an animal must balance, when in 
reality many factors can cause energy budgets to vary 
considerably. The idea of balancing a budget is also 
misleading, because energy budgets are expressed as equations 
and are thus by definition balanced. Because energy budgets 
are neither fixed (to the extent that some intrinsic 
energetic properties are, such as basal metabolic rate) nor 
subject to imbalance, I use the more neutral expressions 
energy management or management of energy budgets. 

General Approach and Goals of this Study 

Much evidence suggests that hummingbirds should be 
precise in their energy management and respond rapidly to a 
variety of potentially energetically stressful conditions. 



Nevertheless, to date no study has directly measured energy 
budgets of hummingbirds foraging under conrrolled conditions 
of intruder pressure and food resource states. This study 
focuses on five related determinants of energetic success: 
foraging mode, food processing constraints, competition for 
food, food availability, and food dispersion or density. The 
experiments discussed below evaluated the energetic 
importance of these determinants both singly and in 
combination. Because many other factors can contribute to 
energetic success, I attempted either to hold these factors 
constant or to control for them by using matched-pairs and 
repeated-measures designs whenever possible. 

Foraging Mode 

In successional habitats at the lower elevations in 
Monteverde, Costa Rica, where I conducted this study, the 
hummingbird guild contains approximately 11-15 species over 
the course of a year (Feinsinger 1974, 1976, 1977, 1978, 
1980, H. Tiebout, unpubl . data), representing a wide variety 
of foraging m^odes (Feinsinger and Colwell 1978) . Two modes 
that differ greatly in both behavior and morphology are 
territoriality and low-reward traplining. The following 
description of behavior and morphology of these modes comes 
primarily from Feinsinger (1976) , Feinsinger and Chaplin 
(1975), Feinsinger and Colwell (1978), and Feinsinger et al. 
(1979) . Territorial hummingbirds are typically extremely 



10 



aggressive, defending flower clumps from other hummingbirds, 
both con- and heterospecif ics (Cody 1968, Austin 1970, Murray 
1978, Wolf 1978) , as well as from insects (Boyden 1978, 
Carpenter 1979, Brown et al. 1981, Gill et al. 1982). 
Territorialists have high wing disc loadings relative to the 
many other foraging modes used by short-billed hummingbirds. 
Wing disc loading is the ratio of body mass to the area (a 
disc) swept out by the wings during a single complete beat. 
In territorialists, high wing disc loading results in 
relatively high power output required for hovering and 
forward flight, but may also permit faster wingbeat which may 
enhance acceleration and maneuverability. Low-reward 
trapliners also appear to prefer dense clumps of flowers if 
available, but usually are excluded by resident territorial 
birds. Thus, low-reward trapliners tend to forage among the 
widely dispersed, often nectar-poor flowers not defended by 
territorial birds . Compared to all other foraging modes used 
by short-billed hummingbirds, low-reward trapliners have very 
low wing disc loading, which results in relatively low ' 
required power output during hovering and forward flight. 

From the available species I selected two, one from each 
foraging mode described above, that were as close as possible 
in body size and bill morphology while differing markedly in 
wing disc loading. These two species comprise the "core" of 
the nectar-feeding guild in the successional Monteverde 
habitats (Feinsinger 1976) . Amazllia saucerottei , the 
Steely-vented Hummingbird, is a mid-sized (3.5-4.9 g) 



11 



territorialist, with both sexes defending flower clumps of 
several species (e.g., Inga brenesii, Leguminosae; Hamella 
patens, Rubiacae; and Lobelia laxiflora, Campanulaceae) . 
Amazilia has high wing disc loading (.0335 g-cm~2, Feinsinger 
et al. 1979) and a short (ca. 23 mm), straight bill. This 
species is found at the study site throughout the year, with 
peaks in abundance during March through May and September 
through November (Feinsinger 1976). Chlorostilhon canlvetii, 
the Fork-tailed Emerald, is a somewhat smaller (1.8-3.0 g) 
low-reward trapliner. Both sexes visit many of the same 
flower species as Amazilia, although they are often 
restricted to flowers that are widely dispersed when 
territorial species are defending clumps. In addition, 
Chlorostilhon visits many insect-pollinated flowers and low- 
reward flowers not often visited by other birds (e.g., Rubus 
spp., Rosaceae; Acnistus arhorescens, Solanaceae; and Cuphea 
sp., Lythraceae) . Chlorostilhon has a short (ca. 18 mm), 
straight bill and low wing disc loading (.0262 g-cm-2^ 
Feinsinger et al. 1979). A 28% difference in wing disc 
loading between Amazilia and Chlorostilhon translates into a 
13% difference in the power required for hovering at the 
elevation of Monteverde {Amazilia > Chlorostilhon) (Feinsinger 
et al. 1979). Chlorostilhon is found in Monteverde from 
January through November, with a peak in abundance from April 
through June (Feinsinger 1976) . 

In addition to having similar bill morphologies, 
visiting many of the same flower species, and occurring 



12 



together at the study site for most of each year, the two 
species interact frequently in the field — especially at rich 
nectar sources. Feinsinger (1976) observed that Amazilia won 
99% of aggressive encounters with Chlorostilbon . Considering 
aggressive encounters among all guild members, those by 
Chlorostilbon were primarily interspecific, whereas those by 
Amazilia were mostly intraspecif ic . Feinsinger concluded 
that, in terms of population-level competitive effects {sensu 
MacArthur 1972) , Amazilia had a relatively large impact on 
local numbers of Chlorostilbon, primarily by excluding it 
from rich nectar sources, whereas Chlorostilbon had 
relatively little effect on Amazilia. In addition, Amazilia 
could limit their own local populations through intraspecif ic 
interference competition. 

This pair of hummingbird species is ideal for a 
comparative study of the four additional determinants of 
■foraging profitability presented below. The above 
attributes, coupled with their divergent foraging modes, 
suggest that the two species should differ markedly in their 
energetic responses to these determinants. Amazilia, with 
its high wing disc loading and interference type of 
competition, should experience greater energetic success than 
Chlorostilbon when food sources are rich and defensible. In 
contrast, Chlorostilbon, with its low wing disc loading and 
exploitative m.ode of competition, should fare better than 
Ajnazilia when food sources are limited or widely dispersed. 
In addition, because the generic pair Amazilia-Chlorostilbon, 



. -'•^»-^,-i-ilai_-~>-' . 



13 



with their contrasting yet complementary foraging modes, 
occurs frequently throughout the lowland neotropics (P. 
Feinsinger, pers . comm.), results based on the Monteverde 
pair may be applicable to successional habitats over a wide 
geographic range. 

Physiological Constraints on Rates of Food Processing 

Diamond et al.(1986) and Karasov et al . (1986) suggest 
that hummingbirds in the field may typically be limited in 
their food intake rates because of physiological constraints 
on food processing, primarily digestion and absorption. If 
this is true, then ecological factors that appear 
superficially to limit intake, such as competition, nectar 
volume per flower, and flower density, may have relatively 
little impact on energy budgets. On the other hand, if 
intake rates were generally not constrained physiologically, 
then the ecological factors that affect food acquisition 
might play a much more im^portant role in determining energy 
budgets . 

My first experiment evaluated the relative importance of 
physiological limiits to food intake in the two study species. 
I quantified food passage rates for both species follov;ing ad 
libitum feeding and following single meals of known sizes. I 
then compared my results to theoretical values from the 
literature and estimated potential maximum intake rates for 
each species based on passage rates. 



14 



Competition for Food 

With their high energy requirements, sensitivity to 
energy stress, and diversity of foraging modes, hummingbirds 
are an excellent group in which to investigate the possible 
energetic consequences of competition for food. Unless noted 
otherwise, I use the term "competition" to denote the 
behavioral interactions among individuals using the same food 
resource. These behavioral interactions have been 
categorized into two classes of competition, interference and 
exploitative (Birch 1957, Miller 1967, Case and Gilpin 1974), 
although more recent studies recognize as many as six kinds 
of competition (Schoener 1983). Because field studies have 
reported syntopic hummingbird species with different foraging 
modes utilizing the same flower species (Stiles and Wolf 
1970, Wolf 1970, 1978, Feinsinger 1974,197 6,1978,1980, Stiles 
1978,1980,1981, Montgomerie 1979, Brown and Bowers 1985), 
intra- and interspecific competition for food appears common 
(see especially Feinsinger 1976, Lyon 1976, Wolf et al . 1976, 
Gill 1978) . To date, however, the actual energetic 
consequences of sharing a food resource, if any, have not 
been measured directly in any hummingbird. 

Each of the two foraging modes I studied corresponds 
closely to the two standard classes of competition mentioned 
above. Amazilia, typical of territorial hummingbirds, 
defends rich nectar sources with a variety of behavioral 



15 



patterns {Feinsinger 1976, H. Tiebout pers . obs . ) . Some 
territorial behavior, such as calls and displays from a 
perch, are relatively inexpensive energetically (Ewald and 
Orians 1983) . Many others, however, appear to be as costly 
as foraging or more so, as when the territory resident chases 
an intruder in flight. Because of its tenacious defense of a 
food source, a territorial hummingbird such as Amazilia 
corresponds closely to the ideal interference competitor 
(Feinsinger 1976), or territorial and encounter competitor 
{sensu Schoener 1983) . 

In contrast, low-reward traplining hummingbirds, such as 
Chlorostilbon, typically visit widely dispersed flowers and 
tend to avoid encounters. For example, Chlorostilbon will 
enter Amazilia territories, but Chlorostilbon do not initiate 
aggressive encounters and usually leave flower clumps 
defended by resident Amazilia. Thus, species with a 
traplining foraging mode appear to behave as the ideal 
exploitative competitor (Feinsinger 1976), or consumptive 
competitor (sensu Schoener 1983) . Birds competing by ' 
interference also compete by exploitation, in the sense that 
available food must be divided among participants. In 

addition to lacking interference, exploitative competition i 

may also involve behavior (e.g., food location and timing of i 

visits) that increases food acquisition relative to others I 

using the resource. For example, trapliners may increase 
visit rate to flowers and harvest nectar "pre-emptively," 
thus making a nectar source unprofitable for a potential 



16 



competitor (Gill 1988). According to Feinsinger (1976), 
Chlorostilbon may be such a skillful trapliner of dispersed 
resources that many of the other hummingbird species in the 
guild are forced to bypass these resources. 

The goal of my second experiment was to determine the 
short-term behavioral responses and energetic consequences of 
sharing an easily defensible food source. I tested Amazilia 
and Chlorostilbon in laboratory trials with a single ad 
libitum feeder (an artificial flower containing unlimited 
sucrose solution) under three conditions: solitary birds, 
conspecific pairs, and heterospecif ic pairs. I addressed two 
general questions. (1) Do hummingbird species with different 
foraging modes exhibit different behavioral and energetic 
responses to sharing a defensible food source? (2) Are 
behavioral and energetic responses dependent upon the type of 
pair, conspecific or heterospecif ic? 

Despite the presence of behavioral interactions between 
paired birds that indicate competition, one cannot assume 
that these interactions will have energetic consequences. 
For example, energetically expensive behavior exhibited when 
birds share food (e.g., territorial display flights) might 
also be exhibited by solitary birds or be replaced by other 
behavior equally as expensive (e.g., mate searching or 
courtship display) . Thus, although there may be time budget 
"costs" or other indirect fitness-related costs associated 
with sharing a food source, there might be no net energetic 
costs when compared to solitary birds. 



17 



Food Availability and Dispersion 

In the absence of competition, two components of the 
food base can be of prime energetic importance to foraging 
hummingbirds: food availability and foraging cost. I define 
food availability as the energy intake per flower and 
foraging cost as the total energetic expenditure (on transit 
and handling) per flower visit. These two components 
together determine foraging profitability, often expressed 
either as net energy gain (energy intake - foraging cost) or 
as foraging efficiency (energy intake/foraging cost) . One 
major determinant of foraging cost is foraging flight time. 
At a constant reward per flower, as dispersion (interflower 
distance) increases so does foraging cost, and hence 
profitability decreases. Similarly, at a fixed interflower 
distance, as flower reward decreases, foraging cost increases 
because a bird needs to visit more flowers to get the same 
amount of food. Thus, as food availability decreases, so 
does profitability. Birds foraging under conditions of 
maximal food availability with minimal foraging costs are 
unconstrained by resource conditions and should be able to 
maintain optimal foraging profitability. Conditions that can 
result in decreased profitability subject birds to "foraging 
constraint, " because birds may need to compensate 
energetically for higher foraging costs and/or lower food 



18 



availability in order to maintain energetic success as 
previously defined. 

Hew do hummingbirds respond to noncompetitive conditions 
of foraging constraint, in which flowers either have low 
nectar reward or are widely dispersed, or both? Because 
species with different foraging modes may differ in foraging 
costs and forage most profitably at different levels of food 
availability (Feinsinger and Chaplin 1975, Wolf et al. 1976, 
Feinsinger and Colwell 1978, Feinsinger et al. 1979, Brown 
and Bowers 1985), they may respond differently in both 
behavior and energy management. The goal of my third 
experiment was to examine the short-term responses of 
hummingbirds to experimental conditions that simulated 
natural foraging constraint. I addressed two specific 
questions. (1) What are the behavioral and physiological 
mechanisms by which birds regulate their energy budgets under 
foraging constraint? (2) Are there species-specific 
differences in energy management that relate to foraging 
mode? I tested solitary individuals of each species in a 
large flight cage at two levels of food availability and two 
perch-to-feeder distances. 

Food Dispersion and Competition 

Competition for food can interact with food dispersion 
patterns to affect foraging profitability and overall energy 
budgets. When nectar is limited, competition for food can 



19 



affect energy budgets in several ways. First, intake may be 
affected because available food is divided among competitors. 
Second, foraging costs per unit nectar consum^ed may increase 
because transit costs increase, because birds must visit more 
flowers or visit the same flowers more often (pre-emptive 
harvesting) . Finally, expenditures may increase due to 
interference competition. The nature of these effects 
depends on the types of competitors present and the 
def ensibility of the food source. 

My fifth experiment investigated the behavioral and 
energetic consequences of sharing limited food at different 
patterns of flower dispersion. Using the doubly labeled water 
method (validated in experiment 4, Chapter 6), I tested 
heterospecif ic pairs at two feeder dispersions: clumped 
(easily defensible) and widely dispersed (nor easily 
defensible) . My goal was to test the prediction that 
Amazilia would be relatively more successful energetically at 
clumped feeders, whereas Chlorostilbon would do relatively 
better when feeders were dispersed. 

In Chapter 2 I present the general materials and methods 
used throughout my five experiments. In Chapters 3 through 7 
I present each experiment with (1) a brief introduction 
specific to the topic, (2) specific questions to be answered 
and predictions to be tested, (3) details of methods not 
covered in Chapter 2, (4) results, and (5) a discussion of 
the results as they apply to the experimental conditions 



20 



only. In Chapter 8 I integrate the results of the 
experiments and suggest extrapolations of my findings to 
natural communities. 



CHAPTER 2 
GENERAL MATERIALS AND METHODS 



Hu mmJ-nqhird Capture and Har^ 



All hummingbirds were mist-netted in second-growth 
habitats at Monteverde, Costa Rica (elev. 1300-1500 m) . I 
conducted all experiments indoors on site. I maintained 
birds in a 13 m^ communal aviary on sucrose solution 
supplemented with avian vitamins and Drosophila, ad libitum. 
The aviary sucrose solution was made with locally obtained 
semi-refined table sugar, which was added to tap water from a 
local spring. The solution was 20% sucrose by weight (± 1%) 
as measured by a hand-held optical ref ractometer (American 
Optical® model 10431, precision +0.25% [=0 . 25°Brix] ) . I 
provided birds with fresh water for drinking and bathing, and 
I changed food and water daily. Throughout their captivity, 
birds experienced natural photoperiod (approx. 12L:12D) and 
conditions of temperature and humidity as close to those 
outdoors as possible (indoor mean daytime temperature 21-27 
°C, relative humidity up to 100%) . I marked birds 
individually with enamel paint and colored acetate leg flags 
(Stiles and Wolf 1973) for individual identification. The 
All-University Committee on Care and Use of Laboratory 



21 



22 



Animals, University of Florida, approved all animal care 
procedures . 

Birds stayed in the communal aviary for 1-5 weeks prior 
to being used in experiments, to allow acclimation to 
captivity. I moved birds to individual holding cages 1-3 d 
before each trial. The day before an experiment, birds 
received no Drosophila . Examination of excreta (urine and 
feces com.bined) indicated that 1 d was sufficient time for 
birds to clear all Drosophila from the ga^stro-intestinal (GI) 
tract (Hainsworth 1974) . During experiments, I fed birds 
20.0 ±0.25% sucrose solution made with fully refined table 
sugar purchased locally and gave birds no Drosophila. This 
diet produced clear, colorless excreta that were apparently 
devoid of solids. To avoid carry-over effects, I used each 
bird in only one experiment, except that several birds in the 
DLW validation had also participated in the previous 
experiment. Following each experiment, with an eye-dropper 
of 20% sucrose solution I hand fed each bird prior to its 
release at its site of capture. 

Energy Budgets 

The standard equation for a daily energy budget (after 
Merker and Nagy 1984) is 

I=R^P + U + F, (2.1) 



23 



where I (intake) is energy consumed as food, R (respiration) 
is energy expenditure, P (production) is energy stored or 
lost as mass, and U and F are energy lost in urine and 
feces, respectively. By measuring the refractive index of 
hummingbird excreta, Hainsworth (1974) determined that 
hummingbirds assimilate sucrose m.eals with approximately 100% 
efficiency. He also found that other solutes in excreta that 
could represent energy loss, such as uric acid, were not 
present in appreciable quantities (< 1%) (Hainsworth 1974, H. 
Tiebout pers . obs . ) . Thus, for experimental birds feeding 
only on sucrose solution, I assume no loss of energy in urine 
and feces, hence Eqn . 2.1 simplifies to 

I = R + P . (2.2) 

Violations of this assumption would cause a slight 
overestimation of R, but should not systematically bias the 
results from any one treatment group over others. 

To judge a bird's energetic success requires knowing its 
"preferred" energy budget, because one cannot assume a priori 
the most appropriate levels of intake, expenditure and 
storage. In the case of a nonreproductive hummingbird not 
preparing for miigration, for example, the preferred 24-h 
energy budget should result in a constant body mass from day 
to day (existence metabolism; Kendeigh et al . 1977); thus, a 
successful energy manager should store sufficient energy 
during the day to cover nighttime expenditures. In contrast. 



24 



premigratory or prereproductive birds should prefer energy 
budgets that result in mass gain from day to day. 

The most conservative method I used to determine 
preferred energy budgets was to test birds under conditions 
that allowed them to "choose" their levels of I, R, and P. I 
did this experimentally by offering birds ad libitum food at 
a minimal transit distance with no competition. I then 
compared the preferred energy budgets measured under such 
"control" conditions to budgets tested under various 
experimental conditions (e.g., food limitation, competition) 
with trials run as close in time as possible to the controls. 

In the absence of control energy budgets, I assessed 
energetic success on a relative basis by comparing budgets 
from two or m.ore experimental treatments. I assumed that the 
treatment group (s) with the highest P were the least 
energetically stressed. I justify using P as the best single 
measure of energetic success or stress for three reasons. 
First, P combines intake relative to expenditure in a single 
value. Second, it is the best energetic predictor of how 
long an animal can survive should energy intake be 
interrupted (see below) . Third, there are no circumstances 
under which birds would maintain P rates higher than their 
preferred level; thus, when comparing Ps among different 
treatment groups the highest can be assumed to be the closest 
to the preferred level. 

I present energy budgets in terms of mass-specific rates 
(J-g-l-h-^ or kJ-g-l-d-l) for two reasons. First, daytim.e 



25 



energy storage is the single most important measure of 
energetic success (see above), and maximal storage rates 
scale as a function of body mass to the 1.0 power (Paladino 
1989, see Collins and Paton 1989 for a review) . Thus, a 
simple per gram comparison is sufficient to control for 
differences in body mass. Second, energy expenditure (and 
hence metabolic scope) ultimately depends upon the ability of 
a given unit of metabolically active tissue to perform work, 
and the requirements of that tissue for oxygen, glucose, and 
other necessary materials . Expressing energy terms in mass- 
specific units reflects the organism's ability to meet both 
tissue-level output requirements (expenditure) and supply 
demands (intake) (Hainsworth 1981a, p. 166) . 

Other units used for expressing energy terms are not as 
compatible with the goals of my study. For example, I did 
not want to eliminate metabolic effects due to differences in 
body mass by incorporating scaling relationships for energy 
expenditure (e.g., BMR a mass-S4^ see Nagy 1987) . Just 

because smaller birds are predicted to have higher mass- 
specific metabolic rates is not a justification for removing 
this difference, as would be accomplished by expressing 
energy terms in units that incorporate the scaling 
relationship (e.g., J-g~-54.h-l) . The differences in mass- 
specific metabolic rates that are due to body size reflect 
imiportant differences in surface area-to-volume ratios (Brown 
et al . 1978) and thus have significant ecological 



2 6 



consequences in terms of required intake rates and maximal 
endurance of energy stores (Paladino 1989) . 

Similarly, expressing energy terms on a per diem whole 
organism basis is generally inappropriate for my study. A 
small bird may expend only half the total energy of a large 
bird each day, but this does not mean that the small bird 
requires only half the foraging time of the large bird to 
meet its energy needs . Similarly, two birds might store the 
same amount of energy during a foraging day, but for the 
smaller of the two this storage may be adequate to last the 
night, whereas for the larger it may mean only enough stored 
energy to last a few hours . 

I calculated energy budgets using the DLW method 
(described in Chapter 6) and the standard gravimetric method, 
depending upon the experiment . I present the general 
calculations for the gravimetric method below. 

Energy intake for each time period was calculated as 

I = [ (If) ( [SUC]) (16.48) ]/[ (BM) (T) ] (2.4) 

where I is the rate of energy intake per period ( J • g"-'- • h~^) , 
Ip is the mass (mg) of sucrose solution consumed by a bird 
during the time period, [SUC] is the concentration of sucrose 
solution (0.20 mg sucrose/mg solution), the heat of 
combustion of sucrose is 16.48 J/mg (Weast 1970), BM is the 
bird's mean body mass for a trial (see below), and T is the 
duration of the period (h) . 



27 



I calculated rates of energy storage from differences in 
digesta-free body masses recorded before and after each time 
period. Birds weighed themselves by sitting on a perch that 
was attached to a top-loading electronic balance (±0.001 g) , 
as first described by Carpenter et al . (1983). The balance 
was calibrated each day and tared to the mass of the perch 
before recording body mass. The balance pan was covered by a 
small plastic screen to ensure accurate readings, unaffected 
by wind and excreta. The digital display on the balance had 
a stability indicator light that denoted when a reading was 
reliable. Thus, movements of the bird on the perch did not 
cause uncertainty in mass determinations. Each body mass 
measurem.ent followed a food deprivation period that was 
sufficiently long to allow passage of excess water (see 
Chapter 3 for validation of this procedure) . I assumed that 
all changes in body mass represented losses or gains of fat, 
for both daytime mass changes (following Hainsworth et al . 
1977) and for 24-h mass changes (following Hainsworth et al. 
1981) . Measurements of respiratory quotients (RQs) for 18 
species of roosting hummingbirds (Kruger et al . 1982) support 
this assumption. These workers found that, starting about 10 
min after birds stopped feeding, RQs began to drop from 1.0 
(pure carbohydrate metabolism) to 0.8 (pure fat metabolism). 
Thus, 

P = [ (BMf - BMi ) (39. 9) ]/ [ (BM) (T) ] , (2.5) 



28 



where P (production) is stored energy (J-g-l-h-l), BMp and BM 
are final and initial body mass (g) , respectively, and the 
heat of combustion of fat is (9.5 kcal/g) (4.2 kJ/kcal) = 39.9 
kJ/g (from Schmidt-Nielsen and Schmidt -Nielsen 1952, Schmidt- 
Nielsen 1975, Nagy 1983) . If energy were stored in a form 
other than fat (e.g., glycogen, protein, or blood glucose), 
then my values of P would overestimate energy storage. I had 
no reason, however, to expect that such overestimates would 
systematically bias one treatment group relative to others. 
I calculated energy expenditure from eq. (2.2) 



I 



R = I - P . 



(2.6) 



I also calculated a Profitability Index (PI) as 



PI = I/R, 



(2.7; 



PI is similar to "foraging efficiency" (e.g.. Wolf et al. 
1975) ; but because it includes non-foraging expenditures, PI 
is a better indicator of overall energetic success. I use 
the term "index" here, because intake-to-cost ratios can 
exceed 100%, and thus are not true measures of efficiency 
(e.g., work efficiency, Bligh and Johnson 1973). 



29 



Behavior. 



During each observation period I recorded several 
behavioral variables, depending on the goals of the 
particular experiment and its design. I present most of 
these below, with a description of the criteria I used to 
score them. Exceptions to these criteria are noted in the 
Methods section for each experiment. 

Visits to feeders were tallied for each observation 
period and summed for all periods combined. A visit to a 
feeder was scored whenever a bird probed the feeder aperture 
during a foraging flight (perch-to-f eeder-to-perch) . Often, 
while hovering in front of a feeder, a bird probed the 
aperture several times in rapid succession. To avoid scoring 
each of these probes as separate visits, I only scored 
additional visits to the same feeder (during the same flight) 
when 10 s or more elapsed between consecutive probes, or when 
an agonistic encounter occurred between consecutive probes. 
To facilitate comparisons, feeder visits are usually 
presented as rates (visits/h) . 

Meal sizes (mg) were either recorded directly from an 
electronic balance (supporting each feeder) following each 
visit, or calculated by dividing the total intake per period 
by the number of visits, depending upon the cage set-up being 
used . 

Flight time was measured to the nearest second with a 
hand-held stopwatch (±0.5 s per flight) and, unless noted 



30 



otherwise, included both hovering and forward flight, 
althougn these may entail different energetic costs (Gill 
1985) . The time spent flying is presented as the percentage 
of total time observed. 

When birds were in pairs, I recorded agonistic behavior 
believed to be energetically important (Ewald and Orians 
1983), i.e., interactions in which at least one participant 
was in flight. I defined three types of interactions as 
follows, with the focal bird scored as either an initiator 
(=A, initiated interaction by moving towards cagemate) or a 
receiver (=B, received body contact from A or flew away from 
A) . Initiator and receiver roles never switched during an 
interaction, thus, the initiator always "won" the 
interaction. (1) chases occurred when A flew towards B while 
B flew away from A. (2) Body contact occurred when A flew 
towards B and touched any part of B. when birds flew towards 
'each other and touched (no initiator, occurred very 
infrequently) body contact was not scored. (3) Perch 
displacements occurred when B flew from its perch immediately 
following body contact or a chase initiated by A, or 
following a vocalization by A. A then flew to occupy the 
perch location previously occupied by B. Similar to the 
scoring procedure for visits, multiple interactions during a 
single flight were scored only when at least 10 s elapsed 
between consecutive interactions or when the initiator fed 
between interactions. Categories varied somewhat among 
experiments. Feeder defenses were scored when chases or body 



31 



contact occurred within a predefined distance (specified for 
each experiment) of the feeder. I also recorded agonistic 
behavior that occurred further than the predetermined 
distance from feeder: perch displacements, body contacts, and 
chases. I used feeder defenses as a distinct category of 
interactions, because those encounters nearer the feeders 
were more likely to reflect competition for food per se than 
were encounters further from feeders. All agonistic behavior 
is presented as rates (acts/h) . i did not score encounters 
that appeared agonistic but relatively unimportant 
energetically, e.g., vocalizations not coupled with feeder 
defenses, chases, or contact, or contact made between perched 
birds (bill fencing or body jabs) that did not involve at 
least one bird in flight. 

Frequency of agonistic acts initiated was used to 
determine which bird of a given pair controlled the feeder 
for each period. The bird initiating the most acts was 
scored as winning control of the feeder for that period, with 
the cagemate scored as losing. Ties were scored when both 
cagemates initiated the same frequency of acts. 



Analvse.s and .qt.^t-ic;^^ 



ics 



Before analysis, I tested all dependent variables for 
homogeneity of variance using Bartlett's Test (Systat® STATS 
module, Wilkinson 1987) and then used appropriate 
transformations (e.g., the square root arcsine method for 



32 



percent flight time) to reduce heteroscedast icity if 
necessary (Sokal and Rohlf 1981, p. 417 ff.) . I usually 
tested for treatment effects using 2 or 3-way ANOVAs (Systat® 
MGLH module, Wilkinson 1987), depending on the experiment. 
Whenever possible, each subject was tested in all treatment 
combinations for a given experiment, permitting a repeated- 
measures ANOVA analysis. Because of high inter-individual 
variation for most measures, this design greatly enhanced the 
ability to detect group and treatment effects. 

Post hoc comparisons were made using independent t-tests 
between species and paired t-tests for repeated measures 
within a species. I refrained from using multiple comparison 
procedures that test all possible pairs of means, as they are 
usually not recommended for factorial ANOVA designs (Chew 
1976, Petersen 1977) . Instead, I tested only a small subset 
of cell combinations, based on (a) results of the A.NOVA, (b) 
biological relevance, or (c) extremeness of responses (Chew 
1976), and used Bonferroni's adjustment (a/k, where k is the 

number of pairwise comparisons) for post hoc multiple 
comparisons of means to preserve the experimentwise alpha- 
level (Neter et al . 1985, Wilkinson 1987) . The Bonferroni t- 
test is one of the most conservative post hoc tests in term.s 
of holding the experimentwise type I error rate below nominal 
level and is recommended as appropriate following both 
standard and repeated-m.easures ANOVAs (Day and Quinn 1989) . 
Significance levels reported for factors or their 



33 



interactions are from the ANOVA, whereas those reported for 
inter-cell comparisons are from t-tests. 

Linear relationships between variables were tested using 
the Systat® MGLH module (Wilkinson 1987) and BMDP 6D module 
(Dixon 1985) . I follow the Systat® convention: R = Pearson's 
correlation coefficient (reported as "r" in other sources) . 
I report R for both correlations and regressions, because the 
test for significance for R in a correlation is 

mathematically equivalent to the t-test for the significance 
of b (null hypothesis P = 0) for a regression: the 

standardized regression coefficient (b, for slope) is 
identical to Pearson's correlation coefficient (R) (Wilkinson 
1987, p. MGLH 4; and Sokal and Rohlf 1981, p. 583) . Readers 
interested in the proportion of variance in the dependent 
variable that can be predicted from the independent variable 
for regressions can calculate R^ (also r^ , the coefficient of 
■determination, Zar 1984: p. 271). I set a = 0.05 for all 

tests of significance, and I report results with 0.05 < P < 
0.10 as trends (Siegel 1956, p. 9) when they relate to other 
relevant results. I report values as mean — SD in the text 
and as mean — SE in figures . 

Data Interpretation and Presentation 

I wanted the three ecological experiments (Chapters 4, 
5, and 7) to provide results applicable to natural systems. 
To ensure experimental rigor, however, I necessarily 






34 



concentrated on only a few of many possible factors and 
tested them in a very restricted environment . This created 
the potential problem faced by many laboratory ecologists: 
with what confidence can the experimental results be 
generalized to animals in the wild? Because of the large 
number of natural events and conditions that can influence 
hummingbird energetics in the field, this is a valid concern 
for the system^ I studied. 

To address this concern, I structured my experiments to 
enhance the applicability of my results. Each experiment was 
designed to be both comparative and multifactorial. The 
comparative element (e.g., Amazilia vs. Chlorostilbon, or 
high food vs . low food) allowed me to determine the relative 
magnitude of responses (e.g., Amazilia < Chlorostilbon by 23% 
for energy expenditure) as well as their absolute magnitudes. 
Although the absolute values for that variable might vary 
with ambient temperature in the wild, for example, the 
relative differences have a greater likelihood of persisting. 
The multifactorial approach (e.g., species x interfeeder 
distance x food level) , while making interpretation of 
results more complex, allowed testing for statistical 
interactions between factors. Because many factors operate 
simultaneously to determine the energy budgets of birds in 
the wild, such interactions are undoubtedly important in 
natural systems . 

In order to present data in a manner that both 
emphasizes relative differences in responses between species 



35 



or ecological conditions and clearly illustrates the 
interactions between factors, I employ a graphical format. 
Most figures for the dependent variables display each 
treatment cell, while in the associated text I summarize 
condensed group and treatment means and present the magnitude 
of important differences with their significance levels. 



CHAPTER 3 
FOOD PASSAGE RATES IN HUMMINGBIRDS 



Introduction 

My first experiment had two goals. First, I determined 
if food intake rate in Amazilia and Chlorostilbon is normally 
constrained physiologically by their food processing rates. 
Diamond et al . (1986) and Karasov et al. (1986) have proposed 
this constraint in their model for food passage in the 
hummingbirds Calypte anna and Selasphorus rufus. Second, I 
determined if passage rates in the study species are reliably 
fast enough to measure accurately digesta-free body masses 
for birds following reasonably short periods of food 
deprivation . 

By examining the rates at which food empties from the 
crop and is assimilated in the intestinal tract. Diamond et 
al. (1986) and Karasov et al . (1986) derived the following 
general model of digestive constraints to intake in 
hummingbirds. First, they concluded that the crop volume in 
their study species decreased (following a 100 jll meal) as a 
negative exponential function of time. They then proposed 
that crop emptying time was set by two digestive processes: 
acidification of the meal in the stomach, and absorption of 
sugars in the small intestine. Finally, they suggested that 



36 






37 



these physiological constraints on food passage rates limited 
rates of food intake. Thus, the large amount of time 
hummingbirds spend perching instead of feeding (Stiles 1971, 
Wolf and Hainsworth 1971, Ewald and Carpenter 1978, Hixon et 
al . 1983) may be required for the crop to empty sufficiently 
to accommodate the next meal. 

To determine if my two study species were normally 
limited in their food intake by physiological processes, I 
derived four predictions to evaluate the Karasov et al. 
(1986) general model. If processes occurring in the rest of 
the gut set crop-emptying time, as Karasov et al . proposed, 
then passage rates through the GI tract should reflect 
emptying rates from the crop. Thus, if waste is not stored 
in appreciable amounts, (1) excretion rates should follow a 
negative exponential decline over time. If crop emptying 
follows an exponential decline over timie following a single 
meal, as Karasov et al. suggest, then the rate of food 
passage from the crop may be a positive function of the 
volume of food in the crop. If (a) crop volume per se sets 
crop emptying time and (b) digestive constraints limit crop 
emptying rates, as proposed by Karasov et al . , then (2) 
maximal crop emptying rates (and thus maximal excretion 
rates) should be positively correlated with the size of a 
meal consumed by a bird with an empty crop, but with an upper 
limit imposed by digestive constraints. If crop and GI 
passage rates usually limit food intake rates, as Karasov et 
al . suggest, then (3) maximal excretion rates should 



38 



correspond closely to intake rates of normally active birds, 
because if birds could pass food faster than their intake 
rates they would not be physiologically constrained. 
Finally, if birds must perch to allow crops to empty 
sufficiently for the next meal, then (4) species with intake 
rates closer to their maximal passage rates should spend more 
time perching than those with lower intake rates. I tested 
the first two predictions directly with experiments conducted 
on Amazilia and Chlorostilbon . The third and fourth 
predictions were evaluated indirectly, using data for both 
species on (a) feeding rates of normally active birds in the 
laboratory and (b) perching times in both laboratory and 
field. 

To meet the second goal, I determined how rapidly 
hummingbirds pass the contents of their crops and GI tracts 
after feeding. Birds that are digesta-free should be at a 
relatively "stable mass" (unbiased by excess water mass), 
with additional mass loss at a minimal baseline rate 
determined by metabolic expenditures and evaporative water 
loss (EWL) . If relatively short deprivation periods (DPs) 
can yield reliable stable body masses without undue stress to 
birds, changes in stable body masses over time could be used 
to measure fat production (and hence net energy gain) during 
the daily foraging period. 



39 



Method s. 

I measured rates of food intake, excretion, and the time 
to reach a stable body mass in six Amazilia saucerottel (body 
mass = 3.98 ± 0.55 g) and six Chlorostilbon canivetii (2.12 ± 
0.19 g) . Care and feeding of birds are described in Chapter 
2. 

For each of 15 trials per species I moved birds to 

3 
individual cages 1 m with ad libitum 20.0% sucrose solution 

(but no Drosophlla) for 2 h. I then deprived each bird of 
food for 75 min while recording its body mass from a perch- 
balance (± 0.001 g) every 5 min (the "^ lib " runs) . This 
length of time ensured that the crop and GI tract would be 
empty for some time before a meal was offered. I then 
offered each bird a vial of 20.0% sucrose solution for 60 s. 
To ensure an adequate range of meal sizes across the 15 
trials, I varied the contents of the vial from 100 to 500 jll . 
I weighed each vial before and after feeding, and recorded 
meal size (± 0.001 g) . I again deprived each bird of food 
for another 75 min while recording body masses every 5 min 
(the " sinqle-mieal " runs) . At the end of each trial I fed 
birds ad llhitum on 20% sucrose solution for 10 min and 
returned them to the communal aviary. I conducted the 
daytime trials between 0730 and 1635 h. I used subjects for 
one trial per day, for a total of two or three trials per 
bird . 



40 



To determine if short DPs could also be used to 
obtain stable body masses at the end of the foraging 
day, I ran a second set of trials on six individuals of 
each species just prior to roosting. The protocol was 
as described above, except that after the experimental 
meal (offered at ca . 1800 h, the normal roosting time), 
I turned off the lights, and birds roosted. I recorded 
body masses for a minimum of 100 min . 

Changes in body mass prior to reaching a stable level 
could be due to excretion, EWL, changes in respiratory 
quotient (RQ) , or all of these. Because I recorded body 
masses after meals (while sucrose was being processed) , RQ 
was probably constant at approximately' 1. Birds consumed 
more than sufficient water to compensate for EWL and maintain 
tissue water balance, as evidenced by high rates of water 
excretion (urine and feces combined) . Thus, were this water 
not lost to evaporation, it would be excreted as excess. For 
purposes of analysis, I included EWL with excretion, and 
equated rate of body mass loss with excretion rate (urine and 
feces) . 

Analysis 

I analyzed data from each run individually by nonlinear 
regression (BMDP program PAR, Dixon 1985) to fit parameters 
of the model 

P3 'Wt 
Mwt = PI + P2 • e ^ (3.1) 






41 



where M^t = body mass loss during the 5 min period ending at 
Wt, Wt = time (min) since deprivation began, and P1-P3 = 
unknown parameters solved by iteration. 

In preliminary experiments both species reached stable 
rates of body mass loss in 40-45 min (H. Tiebout, unpub . 
data) . I calculated the baseline rate of mass loss for each 
run as the mean of the last six mass loss measurements (Wt = 
50 to 75 min) . Then I calculated the time at which the decay 
curve (eq. 3.1 for each run) intersected the 95% confidence 
interval for the baseline rate. This point represented the 
end of "che first 5-min measurement period for which the rate 
of mass loss was equivalent to the baseline rate. I 
subtracted 5 min from this time to obtain the time to first 
reach stable body mass. I used the maximum mass loss during 
any 5, 10, and 15 min time block (MAX5, MAXIO, and MAX15, 
respectively) to determine the effects of meal size on 
excretion rate. 

I performed statistical tests between groups using 1-way 
ANOVA (BMDP program 7D; in cases of unequal variances I" used 
the Brown-Forsythe ANOVA — Dixon 1985) . I analyzed data from 
daytime and nighttime trials separately. One daytime ad lib 
run for each species was lost due to feeder malfunction. 



42 



Results 

Fcr botn species daytime excretion rates conformed to 
negative exponential functions. M. lib runs showed this 
response as soon as deprivation began (with one exception) , 
whereas single-meal runs usually showed a brief lag time 
(meal transit time) before a peak followed by exponential 
decline (Fig. 3-1, daytime trials) . For both species, 
excretion rates were highly positively correlated with meal 
size (Table 3-1) . In all cases the coefficient of 
correlation increased with increasing time period. 

For the M lib runs, MAX5 did not differ significantly 
between species (50.1 ±24.3 and 41.6 ± 12.1 mg/5 min, 
Amazilia and Chlorostilhon, respectively) . The distribution 
for Chlorostilhon, however, was strongly truncated on the 
right, with no MAX5 values greater than 56 mg (Fig. 3-2) . 
Within each species there was no significant difference in 
"the maximum MAX5 value for ad lib (118 vs. 56 mg/5 min, 
Amazilia and Chlorostilhon, respectively) compared to single- 
meal runs (125 vs. 55 mg/5 min, Amazilia and Chlorostilhon, 
respectively) . In each case Amazilia exhibited miaximum 
excretion rates over twice as great as did Chlorostilhon 
(P < 0.05) . 

For the daytime single-meal runs, the relationship 
between meal size in mg (MSIZE) and the time (min) after 
deprivation to reach a stable body mass (TSTABLE) fit a 
linear model : 



43 



Amazilia: TSTABLE = . 097 1 (MSIZE) + 5.048 (3.2) 

(R = 0.608, P < 0.05) 

Chlorostilbon: TSTABLE = . 1001 (MSIZE) + 10.34 (3.3) 

(R = 0.709, P < 0.01) . 

Birds fed ad libitum reached stable body masses very quickly 
after deprivation began (17.9 + 9.3 and 24.4 ± 8.2 min, 
Amazilia and Chlorostilbon respectively) . 

Excretion rates for birds roosting after single meals 
fit neither linear nor nonlinear models consistently. Thus, 
it was impossible to calculate TSTABLE for roosting birds. 
Both species exhibited variable responses (Fig. 3-1) . In 
some instances birds passed very little, if any, of their 
preroost meals (Fig. 3-la) . Although body masses 
"stabilized, " most of the water from the last meal was 
retained for over 2 h. In other instances roosting birds 
exhibited intermittent peaks of excretion with no clear 
stabilization (Fig. 3-lb) . Roosting excretion rates were low 
compared to daytime rates (MAX5 values: 34.7 ± 16.0 and 23.3 
± 10.9, Amazilia and Chlorostilbon, respectively) and showed 
no significant correlation with the size of meal eaten prior 
to roosting (Table 3-1) . 



44 

Discussion 
Tests of Preaictions 1 and 2: Patterns of Excretion Ratp-.q 

Daytime excretion rates after deprivation followed a 
negative exponential decline over time, confirming prediction 
1 in a manner consistent with the Diamond et al . (1986) and 
Karasov et al . (1986) model. In that model, digestive 
processes set crop emptying rates, which also conform to a 
negative exponential function (see also Hainsworth 1978) . 
The strong positive correlation observed between meal size 
and maximum excretion rate, which confirms the first part of 
prediction 2, is also consistent with a negative exponential 
model of crop emptying but not with a linear model. This 
correlation, however, is not consistent with the the second 
part of prediction 2, which was derived from the suggestion 
that digestive processes per se set crop emptying time 
(Karasov et al. 1986) . Instead, my results suggest that the 
regulatory mechanism functions in the crop rather than the GI 
tract, because if processes in the GI tract constrained crop 
emptying rates, then there should exist a crop volume 
threshold above which excretion rate could not increase. 
This was not found, either for birds feeding ad libitum (crop 
volume unknown) or for birds following single meals as large 
as 297 |j'.L (Chlorostilbon) and 372 fiL {Amazilia) . This 
maximum meal size for Chlorostilbon is nearly 3 times the 
size of the single meal given by Karasov et al (1986) to 
somewhat larger birds . These researchers proposed that the 



45 



passage of a 100-(iL meal was constrained by digestive 
processes, yet the passage of the 297-jiL meal appeared 
unconstrained physiologically. 

Evaluation of Predictions 3 and 4: 
Constraints on Intake and Foraging Time 

Assuming that maximum rates of body mass loss during 
five-minute periods (MAX5) estimate maximum excretion rates 
accurately, that these rates are sustainable, and that 
assimilation efficiencies are approximately 100% (Hainsworth 
1974), I estimated average and maximum potential rates of 
food intake from the daytime ad lib runs. These maximum 
potential intake rates apply only to hummingbirds under 
normal, non-stressed conditions. Maximum digestion (and 
hence intake) rates could increase above these estimates when 
birds are subjected to increased metabolic demand (L. Gass, 
pers . comm.). Maximum intake rates would decrease if maximum 
digestion rates were not sustainable over time. Using the 
mean MAX5 for each species, and assuming excretion rate 
represents 80.0% of intake rate (for 20.0% sucrose solution), 
over a 12 h foraging day Amazilia and Chlorostilhon would 
consume 9018 and 7488 mg of food, respectively. These 
figures agree closely with experimentally measured ad libitum 
rates of 10126 ± 1686 and 6731 ± 1054 mg/d for Amazilia and 
Chlorostilhon, respectively (see Chapter 4) . Using the 
maximum MAX5 value for each species to calculate maximum 



46 



potential rates of food intake yields estimates of 22500 and 
10080 mg/d for Amazilla and Chlorostilbon, respectively. 

If these latter estimates represent the physiological 
limits to food intake for unstressed birds under conditions 
when processing rates can be maximized (i.e., when birds can 
have either full crops, large meals, or both), then Amazilla 
appears to feed at less than one-half its potential rate 
(10126/22500 = 45%) . Chlorostilbon may be closer to its 
intake limit (67%), as suggested further by its extremely 
truncated distribution of excretion rates during s^ lib runs 
(Fig. 3-2) . Within the range of meal sizes taken, 
Chlorostilbon appeared unable to pass excreta faster than 56 
mg per 5 min . 

If perching time reflects digestive constraints then 
Chlorostilbon, which appears to be feeding at rates up to 67% 
of its physiological maximum, should spend more time perching 
than Amazilla, which typically feeds at less than 50% of its 
maximum potential rate. In my experiments, however, 
Chlorostilbon generally spent less time perching than did 
Amazilla (Chapters 4,5, and 7). This difference probably 
holds true in the wild. To exploit its typical array of 
widely dispersed resources (Feinsinger 1976, and Feinsinger 
and Colwell 1978), Chlorostilbon may spend up to 75% of its 
waking tim^e in foraging flight (P. Feinsinger, pers . comm.) . 
This is consistent with flight times reported for 
Chloroszilbon in Chapter 5. With such high flight times, 
birds may actually need increased food intake rates to 



47 



compensate increased foraging costs and thus. would be even 
closer to their physiological limits of nectar processing 
than when food is more easily obtained. Because of this 
positive relationship between flight time and food demand, 
the relationship between sitting time and physiological 
constraints on feeding rates seems tenuous at best. Perhaps 
some other process, such as minimizing energy expenditure 
(Schuchmann et al. 1979), should be examined to explain 
perching time in hummingbirds. 

Application to Experimental Design 

To use short deprivation periods (DPs) to obtain 
reliable stable body masses, the DPs must be long enough to 
allow passage of excess water, yet short enough so that birds 
do not incur an irreversible energy debt (see Tooze and Gass 
1985) . Hainsworth and Wolf (1972a) found that fluid passes 
through a hummingbird's digestive tract in about 0.5 h. 
Accordingly, Hainsworth et al . (1977, 1981) used 30-min DPs 
to minimize the effects of water exchange on mass 
measurements. During my daytime sA lib runs, both species 
reached stable body masses after a mean deprivation time of 
less than 25 min. After 45 min, however, some birds began to 
show signs of entering daytime torpor (see Tooze and Gass 
1985), and one Chlorostilhon had to be revived by warming and 
hand-feeding. This further supports the use of a 30-min DP 
as a conservative but reliable method. 



>*5---«j'.t.ii'ir.-^. 



48 



From the regression equations for MSIZE vs. TSTABLE, I 
estimated the largest meal each species could consume and 
still reach a stable body mass within 30 min . Potentially, 
in 30 min Amazilia could process a meal of about 257 mg of 
20% sucrose solution (eq. 3.2), compared to its mean meal 
size of 112.1 ± 42.0 mg (the solitary control days from 
Chapter 4). Chlorostilbon could process about 195 mg (eq. 
3.3), compared to its normal meal of 69.5 ± 19.8 mg (the 
solitary control days from Chapter 4) . If crop contents 
equal the average meal size (Karasov et al . 1986), or even if 
they exceed tv;ice the average meal size, these figures 
provide independent theoretical support that stable body mass 
should be reached within 30 min after deprivation. 

Finally, because passage rate is a negative exponential 
function of time, small errors in estimating the actual time 
to reach stable body mass will result in relatively small 
errors in estimating true stable body mass. For example, one 
Amazilia during daytime deprivation after ad libitum feeding 
reached stable body mass in 15 min, after losing 74 mg. The 
time to lose one-half of this was only 4.7 min. This agrees 
closely with the figure reported by Karasov et al . (1986) and 
Diamond et al. (1986), based on a negative exponential model, 
of 4 . 1 min to empty one-half of a 100-)Il meal of 20% sucrose 
solution from the crop (a 100 |Il meal consists of 86.4 mg 
H2O) . 

Roosting Amazilia and Chlorostilbon processed crop 
contents differently from active birds. For these species 



49 



reliable stable body masses could not be obtained using 
standard 30-min post-roost DPs (or any other duration under 2 
h) , as was done for Archilochus alexandri (Hainsworth et al. 
1977) and Eugenes fulgens and Lampornis clemenciae 
(Hainsworth et al . 1981) . The irregular mass loss curves 
(Fig. 3-1) could reflect changes in crop emptying rates, 
digestion and assimilation rates, waste-water storage and 
excretion rates, and substrates metabolized. For example, 
roosting birds might exhibit reduced crop emptying rates if 
they metabolize carbohydrates directly rather than convert 
sugars to fat that is then metabolized. Because energy is 
lost when sugars are stored as fat, the former would be 
energetically more efficient. Kruger et al . (1982), however, 
found that for 17 species of hummingbirds, RQ changed from 
daytime values of 1.0 (carbohydrate metabolism) to 0.8 (fat 
metabolism) within 40 min after roosting. Roosting birds may 
process meals in essentially the same way as active birds, 
except that waste water may be retained longer. This might 
serve to protect against excessive evaporative water loss 
during "che night . 

Conclusions 

From this experiment I conclude that 30-40 min periods 
of food deprivation are adequate to ensure that Amazilia and 
Chlorostilbon are digesta-f ree . Thus, changes in their body 
m^asses over time are not attributable to excess water and can 



50 



be assumed to represent somatic energy stores such as fat 
(see Ha:-nsworth et al . 1977, 1981). I also conclude that at 
normal rates of feeding, even when birds take large meals 
following long food deprivation periods, food intake rates 
are probably not limited by digestive processes. 
Chloroszilbon, however, may be closer to reaching its 
physiological limits than is Amazilia . 



51 



Table 3-1. Correlations (R) between meal size and excretion 
rate for Chlorostilbon canlvetll and Amazilia saucerottei 
during daytime trials. ** = P < 0.01. 



Correlation between meal size 
and maximum excretion in: 
5 min 10 min 15 min 
(MAX5) (MAXIO) (MAX15) 



Daytime 

Amazilia .9244 ** .9574 ** .9576 ** 

Chlorostilbon .6915 ** .8448 ** .8672 ** 

Combined .8417 ** .8988 ** .9246 ** 



52 



E 

"D 
O 

^- 
<U 

a. 

c 

ii 



A. Amazilia saucerottei 

AS08-. Daytime mealsize 204 mg 
AS05: Pre-roost meaisize 24 4 mg 





Pnr^rnpnpnpnp 



8. Chlorostilbon canivefii 

CC05: Daytime meolsize 167 mg 
CC07: Pre-roost mealsize 191 mg 



p n p n p n 



n pi n p 



10 20 30 40 50 60 70 80 90 iOO ilO 120 130 
Duration of Food Deprivation (min) 



Figure 3-1. Representative rates of body mass loss following 
a single meal for active versus roosting birds. Active birds 
during the day exhibit a characteristic pattern of a short 
lag time before a sharp peak followed by exponential decline. 
Roosting birds exhibited a range of responses, from virtually 
no mass loss following a meal (A) , to very intermittent and 
irregular mass loss (B) . Daytime birds had smaller meals, 
but passed food much more rapidly during the first hour than 
did the roostincr birds . 



53 




■ Ad Lib runs 

ii Single Meal runs 

Amazilia saucerottei 
(N = 29) 



U 



0-8 16 24 32 40 48 56 64 72 80 88 96 104112120128 



6 1 



(ii 

c 

a: 



0) 

E 

3 




Chlorostilbon canivetii 
(N = 29) 



0-8 16 24 32 40 48 56 64 72 80 88 96 104112120128 

Maximum body mass loss in 5 min. (mg) 

[MAX5} 

Figure 3-2. Maximum body mass loss in 5 min (MAX5) following 
ad libitum feeding and following single meals during daytime 
runs. Note that the distribution for Amazilia is 
approximately normal, while that for Chlorostilbon is sharply 
truncated at 56 mg . 



CHAPTER 4 

INTER- AND INTRASPECIFIC COMPETITION 

FOR A DEFENSIBLE AD LIBITUM FOOD SOURCE 



Introduct. -i nn 
Overview 

Few studies have successfully measured the energetic 
costs and benefits of competition in animals. The metabolic 
expenditures of interference competition have been directly 
determined in several ectotherms (Bennett and Houck 1983, 
Jaeger et al . 1983, Riechert 1988). In addition, Feder and 
Arnold (1982) reported high rates of energy expenditure in 
salamanders chased by snakes, a situation somewhat analogous 
to interference competition. No similar measures of the 
costs of exploitative competition are available. No data are 
available on energetic costs or benefits for either type of 
competition for endotherms, although Roskaft et al. (1986) 
and Hogstad (1987) report high metabolic rates for individual 
Parus major that have high dominance rank in their feeding 
flocks, and Malloy and Herreid (1979) report elevated 
metabolic rates for fighting Mus musculus. 

Hummingbirds provide an ideal group in which to 
investigate the energetics of competition for food because of 
their sensitivity to energy stress (discussed above) and two 



54 



55 



aspects of their foraging. First, hummingbirds can be tested 
on a homogeneous diet of known energy content : sucrose 
solution. If the mass or volume of a meal can be accurately 
measured, then its energy content can be precisely quantified 
(see Bolten et al . 1979 for methodology). Second, 
hummingbirds consume food in discrete meals, without 
selectivity (removing preferred items from within a single 
food source) or caching, and thus the exact amount eaten can 
easily be determined. When artificial feeders are oriented 
vertically with a feeding aperture on top, birds rarely spill 
sucrose solution. In my studies, close examination of birds 
using such feeders established that little if any sucrose 
solution adhered to bills, as was confirmed by the near 
absence of bill wiping following meals . Because of these two 
characteristics, when birds shared a single feeder, I could 
determine energy intake for each bird accurately using 
gravimetric techniques. Food competition experiments using 
gravimetric techniques are difficult to perform with animals 
other than nectarivores, because spillage, selectivity, and 
caching make it difficult to determine intake for more than 
one animal per cage. 

Goals and Predictions 

Feinsinger (1976) observed that Chlorostilbon often 
foraged at dispersed flowers when it was excluded from rich 
f lower sources by Amazilia . He concluded that interference 



56 



by Amazilia at rich resources (a) had major competitive 
effects {sensu MacArthur 1972) on local populations of 
Chlorostilbon and (b) served to limit the population size of 
Amazilia through intraspecif ic competitive effects. He also 
noted that Chlorostilbon had little competitive effect on 
Amazilia . The asymmetries in competitive effects observed by 
Feinsinger, coupled with the differences in behavioral 
interactions used by Amazilia and Chlorostilbon 
(interference vs. exploitative competition, respectively) 
suggest that these species may differ in their short-term 
responses to sharing a food source. To investigate this 
possibility, my second experiment was designed to determine 
the behavioral and energetic responses to sharing an easily 
defensible ad libitum feeder for an entire foraging day. The 
ad libitum feeder simulated a rich natural nectar source, 
such as a large Hamelia bush in full bloom, which may often 
serve as a center of intra- and interspecific behavioral 
interactions (Feinsinger 1976) . 

Testing birds in pairs for an entire day represented 
conditions of competition not normally f ound , in the wild (P. 
Feinsinger, pers . comm., H. Tiebout, pers . obs . ) . For 
example, an intruding Chlorostilbon might attempt to feed at 
a bush defended by Amazilia for only a small part of its 
foraging day. The long experimental trials, though, were 
necessary to allow detection of subtle differences in energy 
budgets, and may provide an approximation of the energetic 






57 



consequences experienced by wild birds during those time 
periods in which they actually share a rich food source. 

The goal of this experiment was to (a) compare species- 
specific responses for Arriazilia and Chlorostilbon that could 
relate to foraging mode, and (b) determine the relative 
effects of intra- vs. interspecific interactions. Based on 
interspecific differences in competitive strategies, I made 
the following predictions for conspecific pairs of Amazilia 
(AA) and Chlorostilbon (CC) and heterospecif ic pairs (AC) 
relative to the performance of each species tested as 
solitary individuals. Recall that Amazilia behaves primarily 
as an interference competitor, whereas Chlorostilbon behaves 
primarily as an exploitative competitor (Chapter 1) . I 
predicted that both species would spend more time flying and 
hovering when in pairs than when alone. This increased 
flight time could result from: (a) the trapliner 
Chlorostilbon comipeting exploitatively by making more 
frequent feeder visits (both CC and AC) (Chapter 1, also Gill 
1988) and flying more due to being chased (AC), and (b) the 
territorial Amazilia competing by interference, hence 
increasing flight time due to chasing and being chased (AC 
and AA) . Because hovering and flying are the most expensive 
activities for a hummingbird, I predicted that birds in pairs 
would have higher energy expenditure than solitary birds. 

Based on their differences in foraging mode and body 
size, I predicted that the energetic responses to sharing a 
feeder would be asymmetrical (Perrson 1985) . In CC pairs, I 



58 



expected, that birds should be able to compensate for 
increased energy expenditure with increased food intake . 
Thus, Chlorostilbon in conspecific pairs should suffer no net 
effect on energy storage. In AA pairs, I expected that 
subordinate birds would have reduced access to the feeder and 
that they would not be able to compensate for increased 
expenditures with increased intake. Thus, Amazllia in 
conspecific pairs should on average have slightly depressed 
rates of energy storage relative to controls . Finally, in AC 
pairs, I expected that Chlorostilbon would not be able to 
increase intake to match expenditures whereas Amazllia would, 
resulting in considerable reduction in energy storage for 
Chlorostilhon but not for Amazllia . Thus, for interspecific 
differences, I predicted that the trapliner Chlorostilbon 
would do better than the average Amazllia in conspecific 
pairs, whereas the territorial Amazllia would do better than 
Chlorostilbon in heterospecif ic pairs. 

In addition to testing these specific predictions, I 
also examined the relationships between behavioral responses 
and energetic consequences for each foraging m.ode to answer 
two questions. (1) Do differences in wing disc loading 
affect energy budgets? (2) Are birds that win behavioral 
interactions more successful energetically than birds that 
lose? 



59 

Methods 

Experimental Cage and Protocol 

I mist netted hummingbirds between 25 May and 12 August 
1986, and housed them as described in Chapter 2. Three days 
before each trial I moved two subjects to individual 1 m-^ 
holding cages, where they were maintained on the same diet as 
before. One day before each trial, I moved each bird to its 
own side of an experimental cage that was divided into two 
equal halves by an opaque curtain (Fig. 4-1) . It was 
maintained on ad libitum 20.0% sucrose solution without 
vitamins or supplemental Drosophila . This protocol ensured 
that each bird had time to acclimate to the experimental 
environment prior to a trial. 

The experimental design included three factors : species 
{Amazilia or Chlorostilbon) , bird number (1 or 2), and pair 
type (conspecific or heterospecif ic) , for a total of eight 
cells or treatment combinations. Each trial ran 48 h and 
consisted of a baseline or control day (solitary birds) 
followed by an experimental day (paired birds) . I conducted 
six trials with heterospecif ic pairs and three trials for 
each species in conspecific pairs . Each bird was tested in 
both types of pairs, with one exception per species, 
resulting in six replicates of each treatment combination, 
using a total of seven individuals of each species. 
Treatment order was assigned using a stratified random 
schedule, with half the subjects of each species tested first 



60 



in conspecific pairs and the other half tested first in 
heterospecif ic pairs. Partners were randomly assigned. Each 
bird spent 4-6 d between trials on ad libitum 20% sucrose 
solution (with vitamins) and Drosophila . 

On the control day, each bird had its own ad libitum 
20.0% sucrose solution feeder and a perch attached to a top- 
loading electronic balance (±0.001 g) . At the start of the 
experimental day I withdrew the curtain dividing the cage, 
and the birds shared a single ad libitum feeder located 
between their perches and attached to a top-loading balance 

(Fig. 4-1) . Thus, the feeder was easily visible and 
defensible from either perch. The protocol for each of the 
two days was identical. At 0500 I recorded each bird's 
initial mass. At 0525 I turned on the lights, and birds 
awoke immediately. At 0530 I opened the feeder(s), and birds 
began to feed. I closed feeders for 30 min four times during 
the day (0730, 1000, 1230, and 1500) and recorded digesta- 
free body masses at the end of each food deprivation period 

(see Chapter 3) . At the end of each trial I turned lights 
off at 1725. Birds usually continued to feed under natural 
light conditions for another 15-20 min until they roosted. 

De pendent Variables 

During each of the five feeding periods (0530-0730, 
0800-1000, 1030-1230, 1300-1500, and 1530-roost, periods 1-5, 
respectively) I recorded visits to feeders (1-h observation 



61 



period) and time spent hovering (40-inin obs . per.) for each 
bird. Eecause the cage was relatively small, I assumed that 
all time spent in the air represented hovering, rather than a 
mixture of hovering and forward flight (as for Wolf et al . 
1975) . I scored a visit only when a bird obtained a 
measurable meal (at least 1 mg) and did not include bill 
probes resulting in no intake. I present visit data in terms 
of interbout time: the time in minutes between scored visits. 
On control days, mean meal size per period was calculated by 
dividing food intake (mg; see below) by the number of visits. 
On experimcental days, meal sizes were measured directly as 
taken (see below) . 

When birds were in pairs, I also recorded two categories 
of aggressive behavior believed to be important 
energetically: feeder defenses and agonistic encounters 
(defined in Chapter 2) . Feeder defenses were scored only 
within 25 cm of the feeder and were recorded during each of 
the 1-h observation periods in which I recorded feeder 
visits. During the 40-min observation periods in which I 
recorded hover time, I also recorded agonistic encounters. 
These included feeder defenses as well as interactions 
further than 25 cm from feeder: perch displacements, chases, 
and body contacts. I also calculated total feeder defenses 
and total agonistic acts by summing the initiator and 
receiver scores for each bird per observation period and for 
each experimental day. These totals, which were the same for 
both cagemates, represented the total number of interactions 



62 



exhibited by a pair. I also recorded whether each bird won, 
lost, or tied for control of the feeder for each period 
(defined in Chapter 2) . 

On control days I calculated food intake (mg) by period 
from preweighed Nalgene® vials, with a correction for 
evaporation based on additional standards (food vials kept 
near ca.ges during each trial) 

If = FMj - FMp - (VMj - VMj.) . (4.1) 

where Ip is sucrose solution intake (mg) per period, FMj. and 
FMj, are initial and final feeder vial masses, respectively, 
VMj and VMp are initial and final standard vial masses, 

respectively. On experimental days I calculated food intake 
by period by summing the masses of individual meals, recorded 
from the balance following each visit, as 



^Mm^ (4.2; 

i =1 



where M^.^ is the mass (mg) of the i ^ meal. 

I calculated energy budgets and the Profitability Index 
(PI) gravimetrically (see Chapter 2) . For all terms I used 
the sam.e body mass for a given bird: its initial mass at 0500 
on the experimental day. I calculated energy budgets for 
each of the five feeding periods and for larger time blocks 
as described below. 



63 



Rates of overnight expenditure were calculated from the 
loss of stored energy following equation (2.5), where BMj is 
the initial stable body mass {sensu Chapter 3) following 
roosting, and BM^ is the final body mass (g) , recorded at 0500 
the following morning. 

Analysis 

I analyzed variables for daytime (periods 2-4, 0800- 
1530) or 24-h (0500-0500) blocks. The daytime block started 
and ended with the recording of digesta-free body masses, 
following periods one and four, respectively. Period one was 
not included because subjects may have had especially low 
body water levels when the body mass was recorded at 0500. 
Period five was excluded because roosting birds do not reach 
a reliable digesta-free mass within 30 m.in (Chapter 3) . The 
period 2-4 mass measurements were taken under identical 
conditions (i.e., feeding followed by 30-min food 
deprivation) and ensured comparable calculations of daytime 
energy budgets. For some daytime results involving only 
behavioral measures, and hence not subject to the same 
sources of measurement error, I used all five periods (only 
if so noted) . Twenty-four hour blocks were used to assess 
■the influence of roosting expenditures and possible use of 
nighttime torpor on energy budgets. 

I tested for treatment effects using a 3-way ANOVA with 
species as a grouping factor, pair type as a trials factor. 



64 



and bird number (alone = control day vs. pair = experimental 
day) as repeated measures on a single subject. Other tests 
were as described in Chapter 2 . 

Result?; 

Solitary Birds 

On the average, Amazilia and Chlorostilbon exhibited 
very similar behavior under control conditions (Table 4-1) 
Although Amazilia took significantly larger meals, both 
species had mean meal sizes that were approximately 3% of 
their respective body masses. Both species spent about one- 
fourth to one-third of their daylight hours in the air, 
feeding about every 7 min . For each species, interbout time 
was strongly correlated with meal size, but not with hover 
time (Table 4-2) . There was no significant relationship 
between hover time and meal size. 

Control birds exhibited significant interspecific 
energetic differences. During the daytime, Chlorostilbon 
expended energy and consumed food at greater rates than did 
Amazilia (Table 4-3) . Although expenditure rate varied over 
two-fold among individuals, birds were able to maintain 
daytime intake rates at or above rates of expenditure (Fig. 
4-2) . As a consequence, energy storage rates (Table 4-3) and 
profitability indices (Table 4-4) indicate that both species 
maintained equivalent net profits during the day and exactly 



65 



maintained body mass over 24 h. Both speciea had similar 
overnight expenditure rates and apparently did not enter 
torpor (Table 4-4) . 

Of the behavioral measures, hover time was the most 
strongly correlated with energetic responses, failing only to 
be a significant predictor of energy storage for Amazllia 
(Table 4-5) . The regressions of daytime expenditure against 
hover time for each species (Table 4-5) were used to test 
differences between resting metabolic rates (RMRs = Y- 
intercepts) and incremental flight costs (slopes), following 
Birt-Friesen (1989) . While both species had identical 
incremental flight costs (ca. 503 J • g~'^ -h'^) , Chlorostllbon 
had a significantly higher mean RMR (F-ratio = 12.42, P < 
0.005) . 

Paired Birds 

Behavior . There were no significant treatment effects 
of pair type on hover time (Fig. 4-3) . Across all treatments 
there was a trend for Chlorostilbon to hover more than 
Amazilia (37.4 vs. 27.4% respectively, P = 0.057). There was 
also a trend for both species to hover m,ore in pairs than 
when alone (39.4 ±27.3 vs. 31.6 ±22.5% respectively, 
P = 0.063, N = 24/group) . The single anomaly was Air,azilia 
in conspecific pairs. Unlike other pairs, this group did not 
increase hover time relative to controls and spent 
significantly less time hovering than did Chlorostilbon in 



66 



conspecific pairs (18.2 vs. 50.2%, P < 0.05, Amazilia and 
Chlorostilhon, respectively) . Among paired birds, the time 
spent hovering varied an order of magnitude. The individual 
with the lowest mean hover time was an Amazilia in a 
conspecific pair (8.0%), while the highest mean hover time 
was by a Chlorostilhon in a heterospecif ic pair (84.1 %) . 

Meal size differed by species within each treatment, 
with Chlorostilhon taking significantly smaller meals than 
Amazilia (Fig. 4-4) . The interaction between bird number 
(one or two) and pair type (conspecific or heterospecif ic) 
showed a trend (P = 0.088). This indicates that, relative to 
controls, each species increased meal size somewhat in 
heterospecif ic pairs, but decreased meal size slightly in 
conspecific pairs. Individuals took the most extreme mean 
meal sizes during heterospecif ic trials, with an Amazilia 
taking the largest (370.5 mg) and a Chlorostilhon the 
smallest (31.1 mg) . The single largest meal consumed by an 
individual in each experimental treatment group (Table 4-6) 
was always at, or in excess of, the theoretical maximum for 
that size hummingbird (based on crop volume, see Hainsworth 
and Wolf 1972a for formula) . 

Interbout time showed a significant three-way 
interaction among all factors (species x bird numier x pair 

type; P < 0.05; Fig. 4-5). Relative to controls, Asaazilia 
increased feeding frequency slightly in heterospecif ic pairs, 
but decreased frequency in conspecific pairs . Chlorostilhon 
exhibited the opposite response, resulting in a considerable 



67 



range in mean interbout times for Chlorostilbon in pairs, 
from 4.6 ±1.6 min (conspecific pairs) to 16.7 ±16.4 min 
(heterospecific pairs) . Interbout time was significantly 
positively correlated with meal size for Chlorostilbon but 
not for Amazilia (R = 0.883 vs. 0.049, respectively; for both 
pair types combined, N = 12/sp.) . 

Ability to control the feeder differed by species and 
was significantly affected by pair type (Fig. 4-6) . In 
conspecific pairs, Amazilia tied for control of the feeder as 
frequently as they won or lost. In contrast, Chlorostilbon 
almost never tied in conspecific pairs. In heterospecific 
pairs, Amazilia won every period over Chlorostilbon, with the 
exception of one tie in which each bird initiated one 
agonist ic act . 

Frequency of initiation of both feeder defenses and 
agonistic acts exhibited the same pattern (Fig. 4-7) . For 
both measures, Amazilia was significantly more aggressive 
than Chlorostilbon (e.g., 30.2 ±59.5 vs. 3.6 ±10.4 agonistic 
acts/h, respectively) . Both species showed relatively low 
levels of aggression when in conspecific pairs (e.g., 9.7 
agonistic acts initiated per h, mean of both species) . When 
in heterospecific pairs, however, Amazilia increased its 
aggression about 4-fold, whereas Chlorostilbon dropped its 
aggression levels to nearly zero. Amazilia was capable of 
short bursts of extremely high rates of aggression. The 
maximum rates per period recorded for individual Amazilia 
were 156 and 231 agonistic acts/h, for conspecific and 



68 



heterospecific pairs, respectively. Despite its relatively 
low mean levels of agonistic acts, Chlorostilhon was also 
capable of occasional periods of intense aggression, 
particularly in conspecific pairs. Single bird extremes were 
98 and 48 agonistic acts/h for conspecific and heterospecific 
pairs, respectively. 

I examined the relationship between hover time and each 
of the measures of aggressive behavior: feeder defenses and 
agonistic acts (both initiated and received for each) , as 
well as total feeder defenses and total agonistic acts 
(initiated and received summed for each) . I correlated hover 
time with each m.easure for all periods (1-5) of either pair 
type (N = 60 periods per species) . None of the aggression 
measures was significantly correlated with hover time. 
Nevertheless, che relationship between aggression and hover 
time followed a similar pattern for all measures, as 
presented for total defense and total agonistic acts in Fig. 
4-8. At very low levels of aggression, hover times ranged 
from zero to the maximum (which varied from 80 - 95%) . As 
total aggression increased, both the mean and range of hover 
time decreased, such that at the highest aggression levels 
hover times were actually the lowest and least variable. 

Energetics . As a group, birds of both species 
significantly increased energy expenditure when in pairs, 
without distinction between pair types (Fig. 4-9) 
Chlorostilhon maintained daytime rates of energy expenditure 
greater than Amazilia for all treatment comJoinations (665.6 



69 



vs. 463.6 J-g-i-h-i, P < 0.001). The difference between 
species for controls (156.4 J-g-l-h"!) was increased in pairs 
(247.7 J-g~l-h~-^), because Amazilia increased expenditure in 
pairs only 6.3%, x<7hereas Chlorostilbon increased 19.8% in 
pairs . 

Species differed in their ability to compensate for 
increased expenditure by increasing food intake (Fig. 4-10). 
Chlorostilbon maintained higher intake rates than Amazilia 
across all treatments (736.4 vs. 578.5 J-g-l-h"!, P < 0.005). 
In addition, there was a highly significant interaction among 
all three factors (species x bird number x pair type; 

P < 0.01). This interaction indicates that Chlorostilbon 
increased intake in conspecific pairs, but decreased intake 
in heterospecific pairs, while Amazilia responded in the 
opposite manner. Compared to control birds (Fig. 4-2), 
paired birds had difficulty maintaining daytime intake above 
expenditure (Fig. 4-11) . One fourth of all paired birds fell 
below the line of intake-expenditure equality, including 
birds with the highest and the lowest daytime energy 
expenditures . ■ " 

The net effect of responses in expenditure and intake 
was that birds in pairs had significantly reduced rates of 
daytime energy storage relative to controls (P < 0.01; Fig. 
4-12) . Amazilia in heterospecific pairs suffered the least 
reduction in daytime energy storage (96.8 ±63.2 vs. 90.1 
±65.6 J-g-i-h-1, control vs. heterospecific pairs, 
respectively, P > 0.05). Based on their daytime energy 



70 



storage rates, paired birds ranked in success, as: Amazilia 
(AC) > .^unazilia (AA) > Chlorostilbon (CC) > Chlorostilhon 
(AC) . 

Paired birds also tended to lose body mass over 24 h, 
whereas control birds did not (24-h storage rates: 
alone > pairs, P < 0.01; Fig. 4-13). Overnight energy 
expenditures did not differ by species, bird number, or pair 
type, but there was a significant interaction between species 
and bird number (P < 0.05) . Amazilia had slightly higher 
overnight expenditures in pairs (174.0 ±65.6 J-g-^-h"^) 
compared to controls (see Table 4-4; P > 0.05), whereas 
Chlorostilbon had significantly lower overnight expenditures 
in pairs (148.1 ±48.3 J-g~l-h~l) compared to controls (see 
Table 4-4, down 18%) . Daytime profitability indices (Fig. 4- 
14) followed the same general pattern as energy storage, with 
a trend of higher profitability for Amazilia comipared to 
Chloroszilhon . 

Hover time was again a good predictor of daytime intake 
and expenditure, but not storage (Table 4-7) . As for 
solitary birds, paired birds significantly increased both 
expenditure and intake with rising hover tim:e. I used an 
analysis of covariance on daytime energy expenditure, using 
species and bird number as factors with hover time as the 
covariate, to test for differences among Y-intercepts (RMRs) 
and slopes (incremental costs of flight) , following Tatner 
and Bryant (1986) and Birt-Friesen et al . (1989). All groups 
had equivalent incremental costs of flight. Chlorostilbon 



71 



had signif icantly higher RMRs than did Amazilia (Table 4-7, 
P < 0.001), with a trend for paired birds to have higher RMRs 
than solitary birds (Tables 4-7 and 4-5, respectively; 
P = 0.1) . 

Relationships between behavior and energetics . Mean 
frequencies of aggressive behavior per trial (see above) were 
poor predictors of energetic responses. Aggressive behavior 
received showed no significant correlations with any of the 
energy variables (I, R, P, and PI) . There was a trend for 
aggression initiated to correlate negatively with both intake 
and expenditure, with only one significant correlation. 
Feeder defenses initiated were negatively correlated with 24- 
h expenditure (periods 1-5, R = -0.417, P < 0.05) but not 
with daytime energy expenditure (Fig. 4-15) 

In contrast, controlling the feeder during a given 
period had marked energetic consequences . Factoring out 
species differences (e.g., Chlorostilhon tended to be 
losers), feeder control significantly affected both energy 
intake and storage. During periods when birds lost control 
of the feeder, intake was curtailed (effect of feeder control 
category: P < 0.01), but expenditures were not reduced (Fig. 
4-16) . As a result, in general, birds that won control of 
the feeder had higher rates of energy storage than did birds 
that tied, which in turn had higher storage rates than birds 
that lost (effect of feeder control: P < 0.005; Fig. 4-17). 
The exception to this trend was that Chlorostilhon that lost 
control of the feeder had relatively high rates of storage. 



72 



In contrast, Amazilia that lost control of the feeder had 
lower rates of storage than any other group (by species and 
feeder control category) . Amazilia only lost control of the 
feeders when in conspecific pairs (6 periods, Fig. 4-6) . 

Asymmetry in responses . Table 4-8 summarizes the 
differences in behavioral and energetic variables between 
solitary and paired birds (both species combined) , 
categorized by their ability to control the feeder. Compared 
to controls (solitary birds) , paired birds that lost control 
of the feeder during a period experienced significant 
increases in meal size, interbout time, and hover time with a 
trend of increasing expenditure. This higher expenditure 
coupled with a slight decrease in intake (nonsignificant) 
resulted in a highly significant reduction in energy storage 
rates for losers compared to controls. Birds that tied for 
feeder control showed similar though moderated responses, 
but without any significant differences (due in part to the 
small number of periods in which birds tied) . In contrast, 
birds that won feeder control exhibited only one significant 
response: increased intake. This was coupled with an even 
larger mean rate of expenditure, however, resulting in a 
slight reduction (nonsignificant) in energy storage relative 
to controls . 



73 



Discussion 

On control days, the two species differed in ways that 
reflect differences in both foraging mode and body size. 
Most significant was that the trapliner Chlorostilhon 
expended energy at a higher mass-specific rate, and hence 
required a greater mass-specific rate of intake, than the 
territorial Amazilia . This was due in part to the 
trapliner 's tendency to hover somewhat more and in part to 
its higher mass-specific resting metabolic rate. The higher 
RMR for the smaller species is expected, based on allometric 
relationships (Lasiewski and Dawson 1967, Nagy 1987, McNab 
1988) . 

Data from the control days indicate that birds of both 
species adjusted well to the experimental cage environment . 
During the foraging day they maintained a high enough 
profitability to store adequate energy for lasting the night 
without losing mass over 24 h and without resorting to 
nighttime torpor. The control birds thus represent a sound 
baseline of energy regulation with which to compare the 
effects of sharing a food source. 

On experimental days, with birds tested in pairs, the 
predictions for both species were generally supported. A 25% 
increase over control days in hover time was associated with 
a 13% increase in daytime energy expenditure for paired birds 
(both species combined) . This increase in expenditure 
contributed to significantly reduced rates of storage for 



74 



paired birds in two ways. First, many birds in pairs had 
reduced access to the feeder (nearly all Chlorostllbon in AC 
pairs and about half of Amazilia in AA pairs) . With 
decreased intake and increased expenditure, storage rates 
necess5-rily dropped. Second, both species may have 
difficulty in meeting high energy expenditure with adequate 
intake. For both control and paired birds, regressions of 
intake vs. expenditure (Figs. 4-2 and 4-11, respectively) 
suggest that above expenditure rates of ca . 800 - 850 
J-g~-^-h~-'- both species will, on average, have intake less 
than expenditure. For solitary birds, this upper limit was 
rarely a problem, with only one individual above 800 
J-g~l-h~l, whereas for paired birds six individuals had 
expenditures above 800 J-g~l-h~^ (one Amazilia and five 
Chlorostilbon) . The explanation for this possible "ceiling" 
on intake is unknown, as both species are capable of rates of 
food intake substantially higher than 850 J-g~^-h~l (see 
below and Chapter 3) . 

The species-specific predictions for paired birds, 'based 
on foraging modes, were largely supported. Chlorostilbon 
responded as miight be expected for an exploitative competitor 
using pre-emptive harvesting. Interbout time decreased 
nearly 30% compared to controls for Chlorostilbon in 
conspecific pairs. In heterospecif ic pairs, however, 
interbout time for the trapliner increased dramatically. 
This increase does not indicate that Chlorostilbon was not 
trying to maintain a high visit rate, only that Amazilia was 



75 



quite successful at reducing visits by the trapliner. A 
measure of the "attempts" Chlorostilhon made to visit the 
feeder is the frequency of feeder defenses initiated by 
Amazilia in AC pairs (Fig. 4-7A) : about once every 90 s, in 
comparison to its control interbout time of one visit every 
5.3 min . Thus, Chlorostilhon apparently greatly increased 
its visit attempts in heterospecif ic pairs compared to 
controls . 

Although the trapliner compensated for its reduced meal 
frequency in AC pairs by taking larger meals, its overall 
intake rate dropped slightly relative to controls. Because 
expenditures increased, Chlorostilhon suffered very low 
energy storage in heterospecif ic pairs (105 J-g~^-h~i below 
controls), as predicted. In conspecific pairs, contrary to 
prediction, Chlorostilhon also suffered reduced energy 
storage (69 J-g~l-h~l below controls), although less so than 
in AC pairs . 

This reduced storage in CC pairs is somewhat enigmatic, 
because there was very little interference at the feeder. 
With an ad lihitum feeder, Chlorostilhon should have been 
able tc consume and process enough food to maintain rates of 
storage equivalent to control days. Based on (a) food 
passage rates for this species (Chapter 3) and on (b) the 
allometric relationship between body mass and maximum food 
intake for endotherms (Kirkwood 1983) , Chlorostilhon has an 
estimated upper limit on intake of 1306 and 1045 J-g~^-h~l, 



76 



respectively, considerably higher than the possible ceiling 
of 800-850 J-g-l-h-l reported above. 

The results for Amazilia confirm most of the predictions 
for an interference competitor. However, Amazilia in AA 
pairs reduced hover time relative to controls, contrary to 
prediction. The AA birds had increased daytime expenditures, 
as predicted, despite flying less. This was due partly to 
increased RMRs and possibly to increased but nonsignificant 
incremental costs of flight when in pairs. As expected, the 
AA birds on average had reduced energy storage, relative to 
controls, due to especially low intake rates for the 
subordinate members of each pair. In comparison, Amazilia in 
AC pairs fared well energetically, maintaining energy storage 
rates equivalent to controls. These AC Amazilia, 
encountering no interference from their trapliner cagemates, 
increased intake by a combination of increased feeding 
frequency and larger meal size. 

The behavioral variables of aggression, when used to 
categorize individuals according to their ability to control 
the feeder in a given observation period, correlated well 
with measures of energetic success. This is illustrated most 
clearly by the effect of feeder control on the profitability 
index (PI, Fig. 4-17B) . For both species, PI was greatest 
for winners and lowest for losers, with Amazilia showing a 
more marked response than Chlorostilbon . Although behavioral 
winners were energetically more successful than behavioral 
losers, both groups paid energetic costs relative to unpaired 



77 



birds. The magnitude of energetic responses to sharing a 
feeder was greatest for Amazilia when in conspecific pairs, 
but greatest for Chlorostilhon when in heterospecif ic pairs . 
This pattern reflects both differences in foraging mode and 
the def ensibility of the feeder. When a food source can be 
easily defended, a bird of either species will suffer 
energetically more during those periods of time it is paired 
with Amazilia than when paired with Chlorostilhon . 

Birds of both species tended to hover less when 
experiencing high levels of aggression (either initiated or 
received) compared to low levels . This seems 
counterintuitive, because birds must hover for aggression to 
be scored. Perhaps when aggression levels are high nearly 
all flights are involved in agonistic encounters. 
Subordinates soon learn that whenever they leave the perch 
they will be chased, and hence fly only to attempt feeder 
visits. Similarly, dominant birds may stay perched to 
maintain watch over the feeder, flying only to repel an 
"intruder" each time the subordinate cagemate leaves its 
perch . 

Even when both wing disc loading and foraging mode are 
known for a hummingbird species, its energetic responses to 
sharing a food source can be difficult to predict accurately, 
even in a controlled environment. First, interspecific 
differences in wing disc loading had no measurable energetic 
consequences, despite the estimated differences in the power 
required for hovering (Chapter 1) . This lack of effects may 



78 



partially be explained by differences in mean crop contents 
between species, which can alter actual wing disc loadings 

(see Hainsworth and Wolf 1975, and discussion in Chapter 5) . 
Second, birds need not strictly adhere to their foraging 
modes: for example, Chlorostilhon was capable of initiating 
moderate interference when in conspecific pairs. Finally, 
predictions based on rules that should enable a given bird to 
regulate energy parameters, such as net energy gain, may not 
take into account the constraints imposed by other birds. 
For example, theories of optimal foraging (DeBenedictis et 
al. 1978, Pyke 1978, 1 981b, 1981c, 1984, Gass and Montgomerie 
1981, Montgomerie et al . 1984) and energy regulation 

(Hainsworth 1978,1981b, Hainsworth and Wolf 1983) have been 
used to formulate rules for optimal meal size, meal 
frequency, and patch residence time. Hummingbirds foraging 
under natural conditions might be able to "follow" such rules 
when competition is absent or when it is purely exploitative 
and hence affects only food availability. However, when 
hummingbirds encounter interference competition, which can 
occur even when species are primarily exploitative 
competiuors, subordinate individuals may have little control 
over the size of each meal, its timing, or when to enter or 
leave a patch. 

Foraging rules for subordinates cannot be derived 
empirically by measuring their actual meal sizes or bout 
intervals. For example, in this study a long interbout time 
for Chlorostilbon in AC pairs did not reflect the rate at 



79 



which the trapliner attempted to feed. Conversely, the small 
meals taken by Amazilia in AA pairs do not necessarily 
reflect a "decision" to compete exploitatively . Instead, 
these responses probably result from interference at the 
feeder by an aggressive cagemate . 

These caveats are not intended to suggest that 
hummingbirds do not need to achieve a certain rate of net 
energy gain over the foraging day (cf. "proportional control" 
of Hainsworth 1981b, Hainsworth et al. 1981), or that 
hummingbirds do not follow rules . When a hummingbird faces 
appreciable competition, however, the best rule to follow 
may simply be "get what you can when you can" — a variation on 
the "making the best of a bad job" rule (Hainsworth and Wolf 
1983) . That some individuals of both species, in both pair 
types, took the largest meals they possibly could (Table 6- 
4), rather than a smaller but "optimal" meal, suggests that 
hummingbirds facing competition need not adhere to the 
constraints of optimality. 



80 



Table 4-1. Baseline (control) foraging behavior for solitary 
birds. Data are for all daytime observation periods (1-5) . 
Sample size was 12 days per species, using six individuals 
for two days for each species. Body mass is the mean from 
0500 on day 2. NS = P > 0.05; ** = p < 0.01. 'Difference 
calculated relative to Amazilia . 



Interbout Hover 
Body mass Meal size time time 
(g) (mg) (min) (%) 



Species 



Amazilia 
Chlorostilbon 



4.14 ±.44 
2.34 +.21 



112 ±42 
70 +20 



7.0 ±2.7 
6.5 +1.7 



2 6.1 ±15.1 
31.8 +20.9 



Difference (%) 



- 3i 



7 NS 



+ 22 NS 



Table 4-2. Correlations between behavioral variables for 
Amazilia (A) and Chlorostilbon (C) . BOUT = interbout time 
(min), MEAL = meal size (mg) , HOVER = hover time (%) . 
NS = P > 0.1, T = 0.1>P>0.05, ** = p < 0.01. R denotes 
Pearson's correlation coefficient. 



X 


Y 


Spi 


scies 




Equation 






R 


P 


BOUT 


MEAL 




A 


Y 


= 0.56 


+ 


0, 


.06X 


0.938 


* * 


BOUT 


MEAL 




C 


Y 


= 1.06 


+ 


0, 


.08X 


0.861 


* * 


BOUT 


HOVER 




A 


Y 


= 45.0 


- 


2, 


.7 OX 


0.478 


■ NS 


BOUT 


HOVER 




C 


Y 


= 74.8 


- 


6, 


. 6 6X 


0.554 


T 


MEAL 


HOVER 




A 


Y 


= 37.0 


- 


0, 


. lOX 


0.270 


NS 


MEAL 


HOVER 




C 


Y 


= 56.3 


— 


0, 


.35X 


0.334 


NS 



81 



Table 4-3 . ^Baseline (control) energy budgets (J-g-i-h"!) for 
solitary birds, where I is intake, R is expenditure, and P is 
storage. Data are for daytime (observation periods 2-4) and 
24-h time blocks. Sample size was 12 days per species, using 
six individuals for two days for each species. 
NS = P > 0.05; **= P < 0.01. Difference between Amazilia 
(A) and Chlorostilbon (C) = [ (A-C) /A] x 100%. 





I 


R 


+ 


P 


Daytime : 








Amazilia 


595 ±106 


449 ±123 


+ 


146 ±90 


Chlorostilbon 


720 ±88 


606 ±140 


+ 


114 ±80 


Difference (%) 


+ 21 ** 


+ 35 ** 




- 22 NS 


24-hour: 










Amazilia 


337 ±54 


338 ±59 


+ 


- 1 ±17 


Chlorostilbon 


394 ±41 


387 ±62 


+ 


+ 8 +27 



Difference (%) + 17 ** 



+ 14 NS 



Table 4-4. Baseline (control) Profitability Indices 
(PI = I/R, see Table 4-3 for terms) and overnight energy 
expenditures (J-g-l-h-l) for solitary birds. Data are for 
daytime (observation periods 2-4) and 24-h time blocks. 
Sample size was 12 days per species, using six individuals 
for two days for each species. NS = P > 0.05. Difference 
calculated as in Table 4-3. 



Daytime PI 



24-hour PI 



Overnight 
Expenditure 



Amazilia 
Chlorostilbon 



1.38 ±.32 
1.23 ±.21 



1 .00 ±. 05 
1.03 +.08 



156 ±54 
181 +67 



Difference (%; 



- 11 NS 



+ 3 NS 



+ 16 NS 



Table 4-5. Hover time (%) as a predictor (independent 
variable = X) of baseline (control) daytime energy budgets 
(periods 2-4) for solitary birds. Symbols and sample sizes 
as in Tables 4-2 and 4-3. 





Dependent 




Regression 








Species 


Variable 




Equat 


ion 


R 




P 


A 


Intake 


= 


461 


+ 


5.07X 


0.892 





.000 


C 


Intake 


=s 


634 


+ 


2.34X 


0.683 





.000 


A 


Expenditure 


= 


317 


+ 


5.02X 


0.760 


0, 


.004 


C 


Expenditure 


= 


420 


+ 


5.04X 


0.922 


0, 


.000 


A 


Storage 


= 


144 


+ 


0.05X 


0.010 


0, 


.974 


C 


Storage 


= 


214 


— 


2. VOX 


- 0.868 


0, 


.000 



Table 4-6. Maximum meal sizes for individual hummingbirds of 
Amazilia (AS) and Chlorostilbon (CC) in conspecific (C) and 
heterospecif ic (H) pairs. See Hainsworth and Wolf (1972a) 
for calculation of theoretical crop volume. 







Pair 


Body 


Meal 




% Theoretical 


Species 


ID# 


Type 


Mass (g) 


Volume 


(^11) 


Crop Volume 


AS 


08 


C 


4.465 


545 




100.2 


AS 


05 


H 


4.941 


604 




103.8 


CC 


10 


C 


2.360 


415 




109.8 


CC 


10 


H 


2.455 


475 




123.0 



83 



Table 4-7. Hover time (%) as a predictor (independent - 
variable = X) of experimental daytime energy .budgets (periods 
2-4) for birds tested in pairs (both types combined) . 
Symbols and sample sizes as in Table 4-5. 





Dependent 




Regr 


■ession 






Species 


Variable 




Equation 


R 


P 


A 


Intake 


= 


417 


+ 


5.48X 


0.740 


* * 


C 


Intake 


=31 


528 


+ 


4.31X 


0.795 


* -k 


A 


Expenditure 


= 


326 


+ 


5.72X 


0.867 


* * 


C 


Expenditure 


== 


449 


+ 


5.30X 


0.885 


* * 


A 


Storage 


= 


91 


- 


0.24X 


0.048 


NS 


C 


Storage 


'^ 


79 


— 


l.OOX 


0.295 


NS 



Table 4-8 . Behavioral and energetic correlates of 
controlling the feeder for a period (periods 2-4); Amazilia 
and Chlorostilhon combined (N = 72 periods per species) . See 
text for definitions of lost, tied, and won. Analysis 
performed for each bird during a period on an experimental 
day (X) compared to the same bird during the same period for 
the previous control day (C) using a matched-pairs t-test. 

Difference (A) calculated as: A= X - C. * = P<0.05, 
T = 0.1>P>0.05, NS = P>0.1. 



Lost (N=32) Tied (N=8) Won (N=32) 



Variable 



Meal size (mg) 


+ 39 


* 


-10 


NS 


-3 


NS 


Interbout tim.e (min) 


+ 8 


* * 





NS 


+ 1 


NS 


Hover time (%) 


+ 10 


* 


+ 10 


NS 


+ 5 


NS 


Intake (J-g-^-h-l) 


-36 


NS 


-15 


NS 


+ 34 


* 


Expenditure (") 


+ 65 


T 


+ 57 


NS 


+ 46 


NS 


Storage (") 


-101 


** 


-131 


NS 


-13 


NS 


Profitability index 


-.14 


NS 


-.15 


NS 


-.04 


NS 



84 



Experimental Cage: TOP VIEW 



Day 1 : Solitary birds trial (control) 



2 m 



T 



1 m 



i 




f"^^^^ Opaque"" (obSERVEr) ^^^^' 



balance 



dividing 
curtain 



^ 



balance 



Day 2: Birds in pair trial (experimental) 



n 



X 



o 



^ 



(observer) 



Figure 4-1. Experimental cage. The top cage is Day 1, in 
which birds were tested as solitary individuals separated by 
an opaque curtain. The bottom cage is Day 2, in which the 
curtain was withdrawn, and both birds shared an ad libitum 
feeder. See text for details of protocol. 



y = 396.6 + 0.534X 



R = 0.853 P < 0.01 



til 

5 



> 
(3 

UJ 

z 

LLI 
UJ 



< 

Q 



uuu- 


Chlorostilbon 


^ 


800- 




^^ 


- 


^ 


^ 


600- 
400- 


)^ 


< 

X = Y 


200- 


y/^ Amazilia 




, 


X y = 324.3 


+ 0.602X 


0- 


X R = 0.699 

}■ >- —J 1 1 , 


P < 0.05 
1 ' 1 ' 1 



200 400 600 800 1000 

DAYTIME ENERGY EXPENDITURE 



Figure 4-2. Energy intake (J-g~l-h~l) as a function of 
energy expenditure (J-g~^-h~l) for solitary Amazilia 
(circles) and Chlorostilbon (triangles) during daytime 
periods 2-4 (N = 12 per species) . 



86 



LU 



LLI 
> 

o 



70- 
60- 
50- 
40- 
30- 



20 



10 




Amazilia (A X A) 
Amazilia (AXC) 
Chiorostilbon (C X C) 
Chlorostilbon (AXC) 




ALONE 



PAIRS 



Figure 4-3 . Hover time for Amazilia (A) and Chlorostilbon 
(C) . See text for description of treatments. Pair types are 

conspecific (AXA, CxC) and heterospecif ic (AXC) . 





200- 


N 


160- 


\H 


120- 


< 
UX 


80- 



40 



.^-—'9 




::---41 



.--4^ 




ALONE 



PAIRS 



Figure 4-4. Meal size for Amazilia and Chlorostilbon. 
Symbols and treatments as for Figure 4-3. 



87 



E 


25-1 






m 


20- 


s 




H 


15- 


H 




3 


10- 


o 




m 




cc 


5- 


UJ 




H 
Z 


0-^ 



--A 




ALONE 



PAIRS 



Figure 4-5. Interbout time for Amazllla and Chlorostilhon . 
Symbols and treatments as for Fig. 4-3. 



CO 
Q 

g 

Lll 
CL 



O 

q: 

z: 



20: 

15: 

10- 

5 







Amazilia 
Chlorostilbon 




LOST TIED WON 

CONSPECIFIC 
PAIRS 



i 



i 
i 



i 



LOST TIED WON 

HETEROSPECIFIC 
PAIRS 



Figure 4-6. Frequency of observation periods in which the 
feeder was controlled by Amazilia and Chlorostilbon . Data 
include periods 2-4. See text for description of control 
categories lost, tied, and won. 



88 





DC 


</) 


3 


m 


o 


m 


-r 


-z. 




m 

Li. 
LU 

Q 


LU 


CC 


Q 


111 


LU 


D 


H 


LU 


< 


LU 


H 



60 
50 
40 
30 
20 
10 




^ Amazslia 

M Chlorostilbon 




r 



CONSPECIFIC HETEROSPECIFIC 

PAIR TYPE 








EC 


o 


LU 
Q- 


H 




CO 


O 


z 


yj 


o 


H 


o 


< 


< 


h- 



80 
60 
40 
20 






B 



CONSPECIFIC HETEROSPECIFIC 

PAIR TYPE 



Figure 4-7. Aggressive behavior for Amazilia and 
Chlorostilbon in pairs. (A) Feeder defenses initiated per 
hour, and (B) total agonistic acts initiated per hour. See 
text for behavioral scoring criteria. 



89 



m 



> 
O 



100-1 
80 
60 i 
40 
20 







Amazilia 



!• t •• 



• • 



100 200 300 

TOTAL AGONISTIC ACTS 

PER HOUR 



uj 



100-1 



80-i»» 



60 



Amazilia 



m 
> 
O 

3: 







;{ jT ••• • 

lA • ^ 



B 



100 200 

TOTAL FEEDER DEFENSES 

PER HOUR 



Figure 4-8. Hover time as a correlate of aggressive 
behavior. Total agonistic acts (A) and feeder defenses (B) 
for Amazilia . 



100 



90 



80 



Chlorostllbon 



UJ 

1 

h- 

UJ 
> 

O 

X 







Ih^ 



100 200 300 

TOTAL AGONISTIC ACTS 

PER HOUR 



400 



LU 



DC 
LU 
> 
O 
X 




Chlorostilbon 



100 200 

TOTAL FEEDER DEFENSES 

PER HOUR 



Figure 4-8 — continued. Hover time as a correlate of 
aggressive behavior. Total agonistic acts (C) and feeder 
defenses (D) for Chlorostilbon . 



91 



DC CC 



UJ 


D 


Z 


h- 


ill 






Q 


ni 


Z 


S 


LU 

a. 


h- 


X 


> 


m 


< 




o 





900 1 

800 

700 

600- 

500- 

400- 

300 




ALONE 



PAIRS 



Figure 4-9. Daytime energy expenditure (J-g'^-h"!) 
and treatments as for Fig. 4-3. 



Symbols 



900 1 



>- 




(D 


8U0 


DC 




LU 




Z 
LU LU 


700 


^ 




LU < 




2 ^ 


6on 


r- Z 




1- — 




> 




< 
Q 


500 



400 




ALONE 



PAIRS 



Figure 4-10. Daytime energy intake (J-g~i-h~l). Symbols and 
treatments as for Fig. 4-3. 



92 



1200-1 



lU 1000- 



y = 209.3 + 0.749X (R = 0.829) 
Chlorostilbon 




y = 162.0 + 0.838X 
(R = 0.746) 
-1 — I — I — 1 — 1 — I — I 
200 400 600 800 1000 1200 

DAYTIME 
ENERGY EXPENDITURE 



Figure 4-11. Daytime energy intake (J-g ^-h'^) in paired 
birds as a function of daytime energy expenditure (J-g~^-h"l) 
for Amazilia (circles) and Chlorostilbon (triangles) . 
Regressions for both species are highly significant 
(P < 0.01) . The dashed line (X = Y) denotes equality between 
intake and expenditure . 



,...vr-)- »»S,II— . 



93 



> 

o 
oc 

LJJ 

LU (5 
< 

S O 

< 
Q 



250 



200 



150- 



100- 



50 



0- 



-50 



Amazilia 
-^ — Chlorostilbon 




ALONE 



CONSPECIFIC 
PAIRS 



>- 
O 

cc 
m 
z: LU 

LU o 

< 

LU CC 

2 O 

< 

Q 



150 1 




1 


0-- Amazilia 






■ 




— A - - Chlorostilbon 


125- 
100- 


C 


i 


] 


r 


75- 
50- 






) 




25- 


B "~^^i 


i 


0- 






i-\C _ 




L 



ALONE 



HETEROSPECIFIC 
PAIRS 



Figure 4-12. Daytime energy storage (J-g~^-h~^) . Amazilia 
and Chlorostilbon shown for conspecific pairs (A) and 
heterospecif ic pairs (3) separately. Symbols and treatmients 
as for Fig . 4-3 . 



94 



>■ 

o 

CC LU 
UJ O 

z < 

LU QC 

o 

■7 0) 

<* 

CM 



25 



15- 



-5- 



-15' 



-25- 



-35 




ALONE 



PAIRS 



Figure 4-13. 24-hour energy storage ,( J • g~l • h"^) . Symbols 
and treatments as for Fig. 4-3. 



X 

m 

Q l.6i 



>■ 



< 

LU 

o 

QC 
Q. 



1.4 



1.2- 



1.0 



0.8 




ALONE 



PAIRS 



Figure 4-14. Daytime Profitability Index (PI). See Chapter 
2 for calculation of PI. Symbols and treatments as for 
Fig. 4-3. 



/--i-j-i. ^-j - 



95 



>- IXI 

o cc 

UUI I— 



UJ 



O 



UJ 

31 O. 

CM UJ 



600 



A 

500- 



y = 389.4 - 1 .30x ® Amazilia 

R = ,417 (p<0.05) A Chlorostilbon 

(for both species combined) 

• y = 352.9 - 0.68x 

R = .370 (p > 0.05) 
' 4^ (for Amazilia alone) 




200 



T 1 1 1 1- 

20 40 60 80 100 

FEEDER DEFENSES INITIATED 



(per hour) 







> 


900- 


o . . 




£ ^ 




a DC 




g§ 


700^ 


Q 


/ 


UJ z 


1 


S UJ 


1 


t X 


500 J 



B 



300 



- L 



1 1 1 ■ 1 ' 1 r- 1 ' 1 

20 40 60 80 100 120 

FEEDER DEFENSES INITIATED 

(per hour) 



Figure 4-15. Energy expenditure as a function of feeder 
defenses initiated. (A) 24-hour energy expenditure (J-g" 
^•h~l) and (B) daytime energy expenditure (J-g~^-h~-) for 
both pair types combined. 






Ill 

O Q 
CC ^ 
UJ "J 

IN X 

LU 
UJ 

m ^ 

< UJ 

" < 

z 



1000 
800. 
600 
400 
200 





A ■ C 
LOST 



96 

INTAKE 
EXPENDiTURE 




A C 
TIED 




A C 
WON 



FEEDER CONTROL 



Figure 4-16. Daytime energy intake and expenditure 
(J-g~^'h"^) as a function of feeder control. See text for 
control category descriptions. 



97 



> 

o 

m Lu 

m < 

ir 

m o 

p &) 

>■ 
< 

Q 



150 



50 



-50 



-150- 






7 1 23 9 

H AmazUia 

W Chlorostilbon 



LOST TIED WON 

FEEDER CONTROL 



X 

UJ 
Q 

Z 



LU >- 

< ad* 

O 
QC 
Q. 



2.0 



1.5- 



1.0- 



0.5- 



0.0' 



B 



26 






LOST TIED WON 

FEEDER CONTROL 



Figure 4-17 . Energy storage and profitability as a function 
of controlling the feeder during an observation period. Data 
are for periods 2-4, with both pair types (conspecific and 
heterospecif ic) combined. Numbers by bars are the number of 
periods for each control category (see text for scoring) . 
(A) Daytime energy storage (J-g~^-h~l) . (B) Daytime 
Profitability Index. 



CHAPTER 5 

THE EFFECTS OF FOOD AVAILABILITY AND INTERFLOWER DISTANCE 

UNDER CONDITIONS OF NO COMPETITION 



Introduction 
Background and Goals 

The experiment presented in Chapter 4 corresponds well 
with previous field work on hummingbird energetics that 
focused on territorial birds at rich nectar sources 
(Carpenter and MacMillen 1976a, 1976b, Copenhaver and Ewald 
1980, Baton and Carpenter 1984) . However, territoriality may 
not be representative of most hummingbirds, and to date we 
know virtually nothing about the energetics of aggressive 
birds feeding away from territories nor of the various less 
aggressive species that do not maintain territories. How do 
hummingbirds manage their energy budgets under such 
conditions, in which nectar supplies may be limited and 
distances between flowers may be large? 

The goal of my third experiment was to mimic the 
foraging conditions of birds that face varying food 
availabilities and interflower distances, thereby simulating 
naturally occurring foraging constraint (defined in Chapter 
1) under conditions of no competition. By my definition, 
birds face no foraging constraint when nectar supplies are 



99 



unlimited and transit costs (determined by perch-to-flower 
and interf lower distances) are m.inimal . I consider energy 
budgets obtained under these conditions to be preferred, 
because birds are free to choose levels of I, R, and P (see 
Chapter 2) . In contrast, birds face varying degrees of 
foraging constraint when food is limited or transit costs are 
high, or both. 

I focused on two topics. First, I investigated the 
range of behavioral and energetic flexibility that individual 
birds exhibit in their energy management. Second, I tested 
for interspecific differences in responses to foraging 
constraint that might relate to the foraging mode of each 
species . 

Energy Management 

Although this topic has received little empirical work, 
individual hummingbirds should be capable of considerable 
flexibility in their energy budgets. In natural systems, 
individual hummingbirds often face marked fluctuations in 
food availability (Feinsinger 1980, Stiles 198,0, Montgomerie 
and Gass 1981, Feinsinger et al. 1985), and to survive they 
must be capable of short-term adjustments in the terms of 
their energy budgets. 

To achieve its preferred rate of daytime storage when 
facing foraging constraint, a bird m.ay adjust its time budget 
to compensate for potential changes in intake and 



100 



expenditure. Consider an example in which the number of 
available flowers is fixed. When ecological conditions 
reduce intake rate below the preferred level (e.g., due to 
reduced nectar secretion rates per flower) , a bird may 
respond to conserve energy by reducing the time spent in its 
most energetically expensive activity — flight. If the bird 
waits longer between foraging bouts, flight costs will be 
lowered and foraging profitability (defined as I-R or I/R, 
see Chapter 1) will increase. In contrast, when conditions 
require foraging time to increase (e.g., due to decreased 
flower density, which would increase transit time per 
flower), a bird's overall expenditure will increase if visit 
rate remains constant . Intake should then be increased to 
com.pensate, if possible. If not, then a bird may reduce 
visit rate, reduce its least essential expenditures, or 
metabolize energy stores. Thus, a bird can make many 
possible adjustments in response to foraging constraint, with 
the net effect on energy management dependent upon the degree 
of foraging constraint and the particular combination of 
responses a bird exhibits. 

Experimental Design and Predictions 

I experimentally simulated two components of naturally 
occurring foraging constraint, reduced nectar reward per 
flower and increased distance between flowers, by 
manipulating the delivery rate of sucrose solution and perch- 



101 



to-feeder distance, Tespectively . Individual birds were 
tested in a large flight cage at two levels of food 
availability {ad libitum and ca . 36% below ad libitum, HIGH 
and LOW, respectively) at each of two distances (4 and 20 m 
one-way, NEAR and FAR, respectively) , for a total of four 
treatment cells . The HIGH food NEAR feeder cell represented 
conditions of no foraging constraint, whereas the other three 
cells represented varying conditions of foraging constraint. 

Because birds were tested under conditions of no 
competition, I expected that both species would do better 
energetically when experimental conditions simulated a rich 
flower clump (no foraging constraint) rather than widely 
dispersed or low-reward flowers. Thus, each species should 
fare better when feeders were NEAR compared to FAR and when 
food was HIGH compared to LOW. Based on wing disc loading 
and foraging mode, I made the following general prediction 
about interspecific differences. Because of its lower wing 
disc loading (and presumably lower flight costs) and its 
known ability to exploit widely dispersed and low-reward 
flowers in the field, I expected Chlorostllbon to exhibit 
greater energetic success than Amazllla under conditions of 
foraging constraint. 

In addition to testing these experimental predictions, I 
addressed two questions about general energy management 
abilities. (a) Can birds adjust energy budgets to conserve 
energy at LOW food? (b) Does increasing feeder distance 
necessarily result in increased flight time? 



102 



Materials and Methods 

All hummingbirds were mist-netted between February and 
April 1988 and were housed as described in Chapter 2. 

Flight Cage 

A large flight tube was constructed of translucent 
plastic sheeting (3-5 mil) , and suspended on plastic cables 

(Fig. 5-1) . Design and materials prevented birds from 
perching anywhere except on the single perch provided, v/hich 
was mounted on an electronic balance at one end of the tube. 
I constructed feeders from 12-cc Nalgene® vials with the 
feeding aperture surrounded by a 4-pointed crown to prevent 
perching. In the FAR feeder treatment, the main feeder 

(20.0% sucrose solution) was at the end of the fifth tube 
segment (20 m from perch) , with an auxiliary feeder located 
at the end of the first segment (4 m from perch) . The 
auxiliary feeder provided a minimal reward (5 and 10 |iL of 

30.0% sucrose solution per 20 min for Chlorostilhon and 
Amazilla, respectively) and served to induce- birds to return 
to the perch. In the NEAR feeder treatment, both feeders 
were located at the end of the first segment (4 m from 
perch), which was sealed off from the rest of the tube. In 
both trials the auxiliary feeder provided the same reward and 
was retained as a constant factor for both distance 
treatments . 



103 



In the HIGH food treatment, the main feeder vial rested 
on an electronic balance and was kept filled. Four identical 
standard feeders were filled and run concurrently just 
outside the flight tube to measure evaporative water loss 
(see Chapter 4) . Individual meals were recorded for each 
visit (±0.001 g) and then summed and corrected for 
evaporation for the total main feeder intake per trial. For 
the LOW food treatments, solution was delivered by 
microsyringe autodispensers (Hamilton PB-600®) via 
microtubing to a small nectary (truncated 5-cc syringe 
barrel) contained within the Nalgene® feeder vial. Meals 
were delivered at 10-min intervals such that the main and 
auxiliary feeders combined provided ca . 460 J-g~^-h~-^ for each 
species. For LOW food main feeders and all auxiliary 
feeders, at the end of each trial I measured any solution 
remaining in the feeders (±0.001 g) and subtracted this from 
the total delivered to obtain total intake per trial for each 
feeder. Intakes from main and auxiliary feeders were then 
converted to Joules and summed for total energy intake per 
trial . 

I tested 10 individuals of each species. Mean digesta- 
free body masses (see below) were 4.21 ±0.29 and 2.29 ±0.24 
g, Amazilia and Chlorostilbon, respectively. Two or 3 d 
before its first trial, I ran each bird for a 2 h "training" 



104 



trial in the full flight tube (20 m) to familiarize it with 
the layout. Because the tube had bends but no forks, birds 
easily flew from their perch to the feeder and back again. 
During each training trial I observed a bird until it had 
completed 5 round-trip flights (perch-to-f eeder-to-perch) , 
which I felt was adequate to ensure that it had learned its 
way through the tube. Amazilia and Chlorostilhon did not 
differ in the time each took to learn the tube (x = 81 min 
for both species combined) . 

The day before a trial, I moved each subject to an 
individual 1 m^ holding cage provided with 20.0% sucrose 
solution ad libitum and no Drosophila . On the day of a 
trial, I moved the bird into the flight tube at 0530, where 
it was confined to the area surrounding the perch by a 
mosquito net curtain (Fig. 5-1) . The bird was given ad 
libitum sucrose solution but no Drosophila, and allowed 90 
min to acclimate to the tube and perch. At 0700 I removed 
the feeder and recorded the bird's mass (±0.001 g) . After a 
40-min food deprivation period, I reweighed the bird to 
obtain its initial digesta-free mass (Chapter 3) . At 0740 I 
delivered the first meal and opened the curtain to begin the 
trial. All birds readily found the feeders, regardless of 
distance . 

I recorded two behavior patterns, flight time and number 
of feeder visits, during six 50-min observation periods 
beginning at 0830, 0930, 1030, 1130, 1250, and 1350. At 1530 
I delivered the last meal, and at 1540 the bird was confined 



105 



behind the net curtain and weighed. After a 40-min food 
deprivation period I recorded the bird's final digesta-free 
mass, allowed it to feed ad libitum for 10 min, and then 
returned it to the communal aviary. The total time, from 
initial to final digesta-free masses, was 8.67 h. Each 
subject was run in all four combinations of food and distance 
treatments, with the order randomly assigned. Each bird 
spent 4-9 d between trials in the communal aviary on the 
standard maintenance diet . 

Dependent Variables and Analysis 

Visits to the main and auxiliary feeders were tallied 
for each observation period and summed for all periods 
combined. A visit to a feeder was scored whenever a bird 
probed the aperture during a foraging flight (perch-to- 
■feeder-to-perch) . Additional visits to the same feeder 
(during the same flight) were scored only when 10 s or more 
elapsed between consecutive probes (to main or auxiliary 
feeder) . I recorded flight times for each observation period 
and for all periods summed. I also calculated the minimum 
percent of flight time necessary for making feeder visits as 

[Tm(Vm) + Ta(Va) ] (100%)/Tfl, (5.1) 

where Tm and T^ are the time (s) for roundtrip flights to the 
main and auxiliary feeders, respectively, as determined by 



106 



timing such flights for birds making direct trips (10 and 20 
s for 4 and 20 m perch-to-feeder distances, respectively) . 
Vm and Va are the total visits to the main and auxiliary 
feeders, respectively, and TpL is the total observed flight 
time (s) . I considered additional flight time above this 
minimum as "unnecessary for feeder visits" or "extraneous" 
because birds did not require the extra time to make the 
number of feeder visits they did. Extraneous flight time may 
in fact be used to travel to feeders, or may have other 
important functions, but it is more time than a bird 
absolutely needs to make its given number of visits. 

Energy budgets were calculated gravimetrically (Chapter 
2), using the mean digesta-free body mass ( [initial + f inal] /2) 
for each bird during its trial. Energy intake rate (I) was 
calculated from total energy intake (see above) per 
observation period (HIGH food only) and per total trial. 

I tested dependent variables using a 3-way repeated 
measures ANOVA, with species {Amazilia vs. Chlorostilbon) as 
the grouping variable and food (LOW vs. HIGH) and distance 
levels (NEAR vs. FAR) as trials factors. Other analyses were 
as described in Chapter 2 . 



107 

Result s. 

Behavioral Responses 

Visit rate to the main feeder and to both feeders 
combined (Fig. 5-2) showed significant effects of distance 
(NEAR > FAR, 108 ±117 vs. 44 ±42 visits/h, respectively; P < 
0.05) and food level ■ (LOW » HIGH, 132 ±104 vs. 20.1 ±9.7 
visits/h, respectively; P < 0.01), without significant 
differences between species. Total visit rate varied 11-fold 
from one visit every 3.6 min (HIGH FAR groups) to one visit 
every 18.7 s (LOW NEAR groups), for both species combined. 
The most extreme individuals were a Chlorostilbon (LOW FAR) 
with 20 visits per 300 min (1 visit every 15 min), and a 
Chlorostilbon (LOW NEAR) with 2095 visits per 300 min (1 
visit every 8.6 s) . 

Flight time exhibited a distance trend (FAR > NEAR; P = 
0.078), with a significant interaction between food and 
distance levels (P < 0.05) indicating a stronger effect of 
distance at HIGH food (Fig. 5-3). Chlorostilbon spent' 
consistently more time flying than did Amazilia across all 
treatments (57.0 ±23.5 vs. 41.9 ±19.4%, respectively; P < 
0.05) . Following visit rate, flight time increased 
dramatically at LOW food compared to HIGH (P < 0.01) . Flight 
times varied over 40-fold among individual birds, from 2.3 to 
100% (Chlorostilbon in LOW NEAR and HIGH FAR, respectively) . 



108 



Both species appeared to engage in considerable flight 
time unnecessary for feeder visits (Fig. 5-4) . Birds engaged 
in more of this extraneous flight when feeders were FAR 
compared to NEAR (P < 0.001), with a significant interaction 
between feeder distance and food level (P < 0.001) indicating 
a more pronounced effect of distance at LOW food. Birds also 
had more unnecessary flight time when at HIGH compared to LOW 
food (P < 0.001) . Birds on LOW food needed virtually all of 
their flight time to make visits to the NEAR feeder, whereas 
trips to the FAR feeder accounted for only about 51% of 
flight time. Birds on HIGH food, regardless of distance, 
required only about 20% of their flight time to visit 
feeders. Species did not differ in extraneous flight time. 

Each species exhibited a highly significant negative 
correlation between flight time and the time between visits 
at LOW food {Amazilia: R = -0.644, P < 0.01; and 
Chlorostilbon: R = -0.744, P < 0.01), but no significant 
relationship at HIGH food. Analysis by observation period 
(added as another factor to the ANOVA model) indicated a 
significant effect of period (P < 0.001), with both species 
decreasing flight time over the course of a trial, regardless 
of treatment group. 

Energetic Responses 

All HIGH food groups had equivalent positive rates of 
energy storage of ca . 44 J- g"-"- • h"-'- (Fig. 5-5) , In comparison 



109 



to HIGH food groups, birds on LOW food suffered significantly 
lower (negative) storage rates (P < 0.01), with a trend for 
Chlorostilbon to metabolize energy stores faster than 
Amazilia (P = 0.062) . Feeder distance had no effect on 
energy storage . 

Within each species, birds visiting HIGH food feeders 
consumed food at the same rate at both distances, with 
Chlorostilbon consuming at a significantly higher rate than 
Amazilia (622 ±192 vs. 563 ±135 J-g~i-h"l, respectively; P < 
0.05; Fig 5-6) . For both species combined, the fixed level 
LOW food groups consumed significantly less (ca. 64% of ad 
libitum) than the HIGH food groups (P' < 0.01) . 

Rates of total energy expenditure ■ showed a significant 
response to food level (LOW < HIGH, 628 ±121 vs. 680 ±139 
J-g~^-h~^, respectively; P < 0.05) but not to feeder distance 
(Fig. 5-7) . There was a trend for Chlorostilbon to expend 
more energy than Amazilia in all treatments (mean difference 
= 13.5%; P = 0.072) . Total energy expended showed no 
relationship to visit rate by species. Expenditure was 
directly related to flight time for each species, and both 
species had Y-intercepts significantly different from zero 
(all cells combined, N = 40 per species) : 

Amazilia: R = 412 (±30) + 4.78 (±.65) X Tc^l (5.2) 

[R = 0.767; P < 0.001] , 



and 



110 



Chlorostilbon: R = 479 (±40) + 3.80(±.66) X TpL (5.3) 

[R = 0.684; P < 0.001] , 

where R is total energy expenditure (J-g~^-h~i) and T^l is the 
percent time spent flying. Species did not differ 
significantly in either slope or Y-intercepts (both presented 
as ± SE) . 

I then used an analysis of covariance, with flight time 
as the covariate (=X) , to test for differences in energy 
expenditure (=Y) among treatment groups. The preliminary 
ANCOVA model, to test significance of the flight time x 

expenditure interaction, indicated that all groups had 
homogeneous slopes (P = 0.34) . Treatment means, and thus Y- 
intercepts, were then compared using the standard ANCOVA 
model. Because the Y-intercept is the energy expenditure at 
0% flight time, these values represent the mean resting 
metabolic rates (RMRs) for the treatment groups (following 
Birt-Friesen et al. 1989; see also Chapter 4). There were 
significant effects of distance (NEAR » FAR; P < 0.01) and 
food level (HIGH » LOW; P < 0.001) on RMR, with no 
difference between species (Fig. 5-8). The Amazilla HIGH FAR 
group was anomalous, having the lowest RMR of any group (HIGH 
FAR « HIGH NEAR for Amazilla, P < 0.001) . Mean RMRs 
varied considerably, from a low of 317 ±71.1 J-g~^-h~^ for the 
HIGH FAR Amazilla to 532 ±67.2 J-g-^-h"! for the HIGH NEAR 
Chlorostilbon . 



Ill 



Discussion 
Gener al Responses to Foraging Constraint 

At HIGH food, all groups were equally successful at 
energy storage . Birds on HIGH food levels at FAR feeders 
maintained daytime storage rates equal to those unconstrained 
by feeder distance, with no differences between species. The 
foraging constraint of LOW food resulted in varying degrees 
of energy stress in all groups. LOW food birds lost rather 
than gained energy stores, with the trapliner Chlorostilbon 
sustaining the greatest energy deficit. Again feeder 
distance had no effect . How can this general pattern in 
energy storage be explained in terms of the behavioral and 
physiological factors birds could control? (1) Why did 
feeder distance have no effects? (2) Did LOW food birds 
compensate energetically for restricted intake? (3) Why was 
the trapliner so unsuccessful energetically at LOW food? 

Birds under foraging constraint responded behaviorally 
by increasing foraging expenditures: both visit rate and 
flight time were dramatically higher at LOW food compared to 
HIGH food. Instead of waiting longer between feeding bouts 
to allow sucrose solution to accumulate in feeders during the 
LOW food treatment, birds increased visit rate 6.5-fold and 
increased flight time by 25% over ad libitum food groups. 
Visit rate and flight time did not always respond similarly 
to treatments, however, as evidenced by the treatment effects 



112 



on the flight time needed to make visits (Eqn. 5.1) . For 
example, at HIGH food birds flew 42% more when visiting the 
FAR feeder as compared to the NEAR feeder, yet visit rate 
decreased at FAR feeders. Flight time did not parallel the 
distance treatment effects on visit rate because (a) a single 
visit to a FAR feeder required twice the flight time as a 
NEAR feeder visit, and (b) birds visiting FAR or HIGH food 
feeders spent much more time flying than the minimum required 
for visits. Within the HIGH food groups, birds visiting the 
FAR feeders needed the same percentage of total flight time 
to make visits as did birds visiting NEAR feeders. However, 
because the HIGH FAR birds had higher total flight times than 
the HIGH NEAR birds, the HIGH FAR birds actually spent more 
time making visits. In the LOW NEAR groups, both species 
spent virtually all of their flight time making feeder 
visits . 

High flight expenditures when food was limited explain 
why LOW food birds needed to metabolize energy stores. The 
extreme loss of stored energy by the trapliner Chlorostilbon 
at LOW food was due to its tendency to fly more than 
Amazilia, spending 64.2 ±19.9% of its time flying at LOW 
food. At HIGH food both species flew more to the FAR 
feeders, with Chlorostilbon flying more than Amazilia, yet 
all HIGH food groups maintained equivalent positive storage 
rates. Did HIGH food birds compensate for increased flight 
time expenditures by increasing intake or by reducing other 
expenditures? 



113 



At HIGH food the more active species, Chlorostilbon, 
consumed food at a higher rate than Amazilia, but neither 
species compensated for feeder distance by eating more. 
Instead, both species at the HIGH FAR feeders apparently 
responded by conserving energy in the same manner that birds 
at LOW food did — data in Fig. 5-8 suggest that they reduced 
their resting metabolic rates. This apparent daytime 
hypometabolic compensation (DHC) to foraging constraint was 
not as severe a depression in metabolic rate as either normal 
nighttime torpor (Bartholomew et al . 1957, Lasiewski 1963, 
Hainsworth and Wolf 1970, Carpenter 1974, Hainsworth et al . 
1977) or experimentally induced daytime torpor (Tooze and 
Gass 1985) . Birds exhibiting DHC did not assume the 
characteristic posture, behavior, or piloerection of birds in 
any depth of torpor (Carpenter 1976, Kruger et al . 1982) . 
Instead, birds remained alert and active but apparently 
allowed their RMRs to drop somewhat between flights. 
Perching birds maintained sleek plumage and usually kept 
their eyes open, but often appeared somewhat sluggish and had 
low body temperatures (judged by touch ca . 30-33 °C, vs. 
normal Tb of 41-42 °C) on the two occasions I handled birds 
to check for torpor. 

Several mechanisms may contribute to lowered RMRs for 
birds in the foraging constraint treatments. First, RMRs 
reflect all activities on the perch (including degree of 
alertness, preening, calling, and beating wings), not just 
the energetic costs of sitting per se . A reduction in 



114 



activity while on the perch could reduce RMRs and would be 
especially marked for sleeping birds (see Beuchat et al . 
197 9) . Second, birds in the LOW food treatments might have 
relatively low RMRs because of greater specific dynamic 
action (SDA) in HIGH food groups. Specific dynamic action is 
the increase in metabolism that accompanies digestion and 
assimilation of food (Eckert and Randall 1983) , which could 
increase RMRs for birds in the HIGH food groups because they 
consumed more food (K. Nagy, pers . comm.) . Effects related 
to SDA, however, probably do not contribute to DHC observed 
in the HIGH FAR groups. Third, perching birds may have used 
hypothermia to reduced RMRs, as discussed below. 

DHC has not been previously reported for any endotherm 
solely as a response to foraging constraint as defined here, 
which in my experiment included a treatment with ad libitum 
food availability (HIGH FAR feeders) . Possible nontorpid 
daytime reduction in body temperature (and hence presumably 
RMR) was suggested, however, for several hummingbird species 
as a response to factors other than foraging constraint . 
Selasphorus rufus (Beuchat et al. 1979), Collbri coruscans 
(Schmidt-Marloh and Schuchmann 1980), Collbri delphinl and 
Amazilia tzacatl (Schuchmann and Schmidt-Marloh 1979a) and 
Trochilus scitulus and T. polytmus (Schuchmann and Schmidt- 
Marloh 1979b) exhibited this response to cold ambient 
temperatures. Another species, Oreotrochllus estella, gave 
this response under cold ambient temperature and food 
limitation (Carpenter 1976, p. 57; and pers. comm.) . DHC 



115 



during incubation at a warm ambient temperature was reported 
for Trochilus scitulus (Schuchmann and Jakob 1981) . During 
bouts at the nest these females became hypothermic (and hence 
hypometabolic) at a Ta of 18 °C, allowing Tbs to drop an 
average of 7.7 °C . The birds experienced neither thermal 
stress nor foraging constraint per se, and the authors 
concluded that birds apparently dropped RMRs to reduce the 
amount of time spent off the nest foraging. During periods 
of DHC these birds remained active on the nest and did not 
enter torpor. 

In my experiment, the use of DHC by f oraging-constrained 
birds resulted in significant energy ' savings . Despite 
increased foraging expenditures at LOW' food and FAR feeders, 
total daytime expenditures for LOW food birds were 8% lower 
than for HIGH food birds, with no effect of feeder distance. 
Thus, although both species lost mass at LOW food, these 
losses would have been much greater without DHC. Had LOW 
food groups maintained expenditures per unit flight time 
equal to unconstrained groups (HIGH food NEAR feeder) , rates 
of stored energy loss would have been 92% greater (-168 vs. 
estimated -322 J-g"l-h~^, with and without DHC, respectively) . 

The energetic response of DHC used by these two tropical 
hummingbirds appears analogous to the thermoregulatory 
strategies of several flying nectarivorous heterotherms . 
Large sphingids and Bombus spp . are facultatively 
endothermic, maintaining high body temperatures during 
activity but tracking amibient temperature at other times 



116 



(Heinrich 1970,1971,1974,1975, Heinrich and Bartholomew 1971, 
Bartholomew and Casey 1978) . Heinrich (1977) argued that if 
body warm-up time is short, endothermy between activity bouts 
is energetically disadvantageous for small animals because of 
their relatively high costs of thermoregulation. He noted 
that this holds true only when animals at low Tb are able to 
avoid detection by predators, because reduced Tb impedes the 
ability to detect and escape potential danger (see also 
Heller 1989) . 

Amazllla and Chlorostilbon appear well-suited for short- 
term heterothermy . Because of their small body mass, when 
endothermy is abandoned (or reduced) , TbS will rapidly fall 
to levels at which significant energy savings will be 
realized (Bucher and Chappell 1989) . Because daytime ambient 
temperatures in Monteverde are relatively warm and stable, 
birds will have dependably short warm-up times and hence 
reduced start-up costs (Hammel et al . 1968). If tropical 
hummingbirds do not often use DHC, it may be due primarily to 
the risk of predation. Perhaps only when this risk is 
exceeded by those associated with low ambient temperatures or 
foraging constraints will hummingbirds be likely to allow 
metabolic rates to drop during periods of inactivity. 

Because hummingbirds incur risks when using DHC, I 
conclude that, as expected, both species were most successful 
energetically under conditions of no foraging constraint 

(NEAR HIGH) . This was the only treatment combination for 



117 



which Amazilia and Chlorostilbon maintained positive rates of 
storage without using DHC . 

Interspecific Differences 

Incremental flight costs (extra energy expenditure per 
unit flight time above RMR) did not differ between species or 
with levels of distance and food, as indicated by the 
homogeneity of slopes in the preliminary ANCOVA. Thus, once 
again interspecific differences in wing disc loading did not 
appear as a significant factor in determining energy 
expenditure (see also Chapter 4) . However, the wing disc 
loading values used to estimate power output requirements 
(Chapter 1) are only representative means, whereas the actual 
wing disc loading of an individual varies erratically over 
the foraging day as body fat is gained or lost (up to 200-300 
mg either way) and with crop filling and emptying 
(fluctuations up to 20% of body mass over less than 1 h; 
Hainsworth and Wolf 1972a) . Thus, the actual wing disc 
loading of experimental birds was impossible to calculate, 
and such fluctuations might have changed the relationship in 
wing disc loading from that based on published values . 

Factors other than wing disc loading can affect 
incremental flight costs and could obscure the effects due to 
wing disc loading. For example, energy expenditure depends 
on the speed at which hummingbirds fly, with maximum 
efficiency occurring at an intermediate velocity. Cage 



118 



dimensions and layout may have caused one or .both species to 
fly at suboptimal velocities, compared to those used in the 
wild, thus masking differences due to wing disc loading. 

Consistent with previously described foraging modes, the 
trapliner Chlorostilbon appeared predisposed under all 
conditions to fly more than the territorial Amazilia, 
although the two species did have similar visitation rates. 
Because greater flight time resulted in consistently higher 
rates of expenditure for Chlorostilbon^ the two species 
differed in patterns of energy management in response to 
foraging constraint. Contrary to prediction, the trapliner 
did relatively poorly at LOW food, regardless of distance, 
having higher expenditure rates and lower storage rates than 
the territorial species. Thus, although Chlorostilbon may 
specialize on flowers with low individual nectar reward, this 
low-reward trapliner requires greater overall mass-specific 
intake rates than does Amazilia . 

As expected, when visiting the HIGH food FAR feeder, 
Amazilia experienced greater energy stress than did 
Chlorostilbon . In this treatment, Amazilia had the lowest 
visitation rate but highest flight time for any treatment 
group of that species. As a result, Amazilia HIGH FAR 
exhibited the most extreme DHC and was the only group of 
either species with RMRs lower during ad libitum food than 
during LOW food. This difficulty in energy management by 
Amazilia was caused by its excessive flight time in the HIGH 
FAR treatment. Although birds were tested individually. 



119 



Amazilia spent much time flying near the main feeder without 
visiting and appeared to be checking for intruders or 
guarding the feeder. This resulted in a mean flight time of 
48.2% for the FAR feeder, compared to only 28.0% for the NEAR 
feeder. Amazilia appears to be an ineffective manager of 
expenditures when rich resources are widely dispersed. 



120 



FLIGHT CAGE 



4 m 



SEGMENTS: 

I 



/ 



II 



III 



IV 



V 



FAR 

feeder 

balance 



NEAR 

feeder 

balance 



^ 



PERCH 
BALANCE 

RETENTION 
CURTAIN 



AUXILIARY 
feeder 




MIRRORS 



Figure 5-1. Experimental flight tube. In NEAR feeder trials 
only tube segment V was used (entrance between IV and V 
closed). In FAR feeder trials the bird flew through all five 
segments to reach the main feeder. Observer used mirrors to 
see down segments I and V. See text for description of food 
delivery rates and feeder distances. 



121 



.E 12001 
E 

(/) o 

H " 900- 

C/) 

— 0) 

> E 

■^ 600 ^ 



CO 

p. 



300- 







A. 



NEAR- 
FAR - 


-o-_- 

— #— 


Amazilia 


NEAR - 

FAR 


--A:-- 


Chlorostilbon 








LOW HIGH 

FOOD AVAILABILITY 



Figure 5-2. Visits to main and auxiliary feeders summed for 
all six 50-min observation periods (N = 10 per cell) . The 
territorialist Amazilia (circles) and the low-reward 
trapliner Chlorostilbon (triangles) were run at LOW and HIGH 
food and at NEAR (open symbols) and FAR (solid symbols) 
feeders . See text for description of food and distance 
treatments. Data points staggered to accommodate error bars, 
here and following figures. 



122 



o 


-— ^ 


z 


S^ 






> 




_J 




u. 


O 




z 


z 
m 

Q. 


DC 
UJ 
> 

O 




X 


HI 




?= 


cc 


1- 


o 



70- 



60 



50- 



40- 



30- 



20 




LOW 



HIGH 



FOOD AVAILABILITY 



Figure 5-3. Flight time as a function of food level and 
feeder distance for Amazilia and Chlorostilbon . See Figure 
5-2 for description of treatments and symbols. 



Q 


120-1 


UJ 


. 


Q ^^ 




LU S^ 


100- 


Ijj ^ 


. 


z 


80- 


IME 
SITS 


60- 


^-> 


■ 


h- 


40- 


I DC 


■ 


O O 


20- 


U- 


0-^ 





LOW HIGH 

FOOD AVAILABILITY 



Figure 5-4. Minimum percent flight time needed for visits to 
both main and auxiliary feeders as a function of food level 
and feeder distance for Amazilia and Chlorostilbon. See 
Figure 5-2 for description of symbols. 



123 



HI 

O 

< 

tr 

o 
\- 

> 
O 
CC 
LLI 

z 

LLJ 



100-1 







-100- 



-200 



-300 




LOW HIGH 

FOOD AVAILABILITY 



1 



-1' 



as a 



Figure 5-5. Daytime energy storage rates (J-g' 

function of food level and feeder distance for Amazilia and 

Chlorostilbon . See Figure 5-2 for description of symbols. 





900 


HI 




^ 




< 


800 


h- 




z 






700 


> 




o 




cr 


600 


lU 




z 




Lii 


500 



400 




LOW HIGH 

FOOD AVAILABILITY 



Figure 5-6. Daytime energy intake rates (J-g~l-h~l) as a 
function of food level and feeder distance for Amazilia and 
Chlorostilbon. See Figure 5-2 for description of symbols. 



124 



m 

Q- 

Z) 
H 

Q 
Z 
LU 

a. 

X 

m 



HI 

z 

HI 



800 n 



700- 



600- 



500 




LOW HIGH 

FOOD AVAILABILITY 



Figure 5-7. Daytime energy expenditure rates (J-g"^-h~^) as 
a function of food level and feeder distance for Amazilla and 
Chlorostilbon . See Figure 5-2 for description of symbols. 



LU 

J— 
< 



o 

-J 

o 

CD 

< 
I— 
LU 



O 



LU 

cc 



600 1 



500- 



400- 



300 



200 



.'■A 




LOW HIGH 

FOOD AVAILABILITY 



Figure 5-8. Daytime resting metabolic rates (J-g~l-h~l) as 
a function of food level and feeder distance for Amazilla and 
Chlorostilbon . See Figure 5-2 for description of symbols. 



CHAPTER 6 

VALIDATION OF THE DOUBLY LABELED WATER METHOD (^HHISo) 

FOR MEASURING WATER FLUX AND CO2 PRODUCTION 

IN AMAZILIA SAUCEROTTEI 



Introduction 

The gravimetric method can be used to calculate energy 
budgets for caged birds only under very strict conditions. 
The experimenter must be able to determine both intake and 
storage with a high degree of accuracy for each participant 
in a trial. When birds share one or more feeders, individual 
meals for each bird must be recorded immediately. In 
preliminary experiments I determined that I was unable to 
follow more than one bird at a time when multiple feeders 
were available. Thus, to conduct my final experiment, in 
which two birds shared five feeders, I used the doubly 
labeled water (DLW) method so that I could calculate all 
terms of the energy budgets for both birds in a trial. 

Although many studies of avian energetics have focused 
on the Trochilidae, only two published investigations (Powers 
and Nagy 1988; Weathers and Stiles 1989) employed the DLW 
method to measure energy budgets and water flux in this 
family. The accuracy of the DLW method has been determined 
for several bird species, but prior to this study the method 
had not been validated for a hummingbird. Because validation 

125 



126 



studies done with other species may not generalize to 
hummingbirds, especially those living in tropical habitats, 
the goal of this experiment was to validate the DLW method 
for one of the two study species under environmental 
conditions identical to those under which I would conduct 
experiment 5 (Chapter 7) . 

I conducted a special validation on a tropical 
hummingbird species for several reasons. Preliminary results 
from a validation study on tropical reptiles indicated that 
small animals living in humid environments may show 
appreciable errors in DLW measurements using tritium (K. 
Nagy, pers . obs . ) , presumably due to the physical 
fractionation of isotopes evaporating across highly permeable 
skin. Lifson and McClintock (1966) recognized the problem of 
fractionation, which results when isotopically labeled water 
and CO2 move between liquid and gas phases at slightly 
different rates than do unlabeled compounds, and provided 
equations to correct for these effects in studies 
incorporating deuterium (^h) and -"-^O as tracers. The two 
validation studies done on birds to test the accuracy of 
^HH-^^O have incorporated corrections for fractionation effects 
and have revealed errors that are not statistically 
significant (mean errors = +3.6% for both studies, range of 
individual anim.al errors = -12% to +17%, LeFebvre 1964; Hails 
1979) . The DLW studies done using tritium (^h) instead of ^R 
(Nagy 1980) may be subject to larger fractionation effects 
than those using deuterium (Nagy and Costa 1980) . 



127 



Nevertheless, validation studies on five species of birds, 
involving 3hh18o (HTO-18) with no correction for isotopic 
fractionation, have indicated good accuracy (mean errors 
range from -4.9% to +6.5%, range for individual values from 
-17% to +16%, summarized by Nagy 1989) . However, because 
hummingbirds have extremely high rates of water flux due to 
their nectar diet, fractionation in this group may be greater 
than for other birds . 

In addition, hummingbirds may be prone to greater error 
in DLW measurements due to dynamic biological effects (Lifson 
and McClintock 1966) , in which one or more water compartments 
in the body differ in isotope activity levels (and hence the 
rates of isotope turnover) from that of the average body 
water. Because the liquid meals of hummingbirds have 
extremely rapid transit times (Chapter 3) and glucose 
movement across their intestinal lumen may involve only 
active transport (Diamond et al . 1986; Karasov et al. 1986), 
unlabeled water from food may not mix completely with 
isotopically labeled body water before being excreted. 

Because of these potential sources of error in applying 
the DLW method to hummingbirds, coupled with increasing 
accuracy in m.easuring isotope concentrations that allows 
detection of small errors (Schoeller, Leitch, and Brown 
1986), I conducted a validation study on caged Amazilia under 
conditions similar to its normal tropical environment. The 
experimental design was to determine the accuracy of the DLW 
(HTO-18) method for Amazilia by comparing estimates of water 



128 



influx and energy budgets as determined by two independent 
techniques: the DLW method (using HTO-18 with and without 
fractionation corrections) and the standard gravimetric 
(balance) method I had used in previous experiments with the 
two study species. 

Materials and Methods 
Protocol 

I used 6 Amazilia (mean body mass = 4.14 ±0.37 g) mist 
netted during March and April 1988. Birds were housed and 
tested indoors under conditions of light, temperature, and 
humidity equivalent to those naturally occurring outdoors. 
Prior to the experiment, birds were housed as described in 
Chapter 2 . The use of tritiated water was approved by the 
Radiation Control Officer (Donald L. Munroe) , Environmental 
Health and Safety Division, Nuclear Sciences Center, 
University of Florida, Gainesville, FL, USA. Permission to 
conduct research with radioactive isotopes in Costa Rica was 
granted by the Departamento de Control de Radiaciones 
lonizantes, Ministerio de Salud, San Jose, Costa Rica. 

One d prior to each trial I moved the subject to an 
individual 1 m^ cage with ad libitum 20.0% sucrose solution, 
but no Drosophila . For each validation trial I weighed the 
bird (±0.001 g) at ca . 0530 (before its first meal of the 
day) and injected it intramuscularly with 50.0 |IL of water 



129 



containing 97 atom % ^^0 and 38 fiCi tritium. Each bird then 
fed ad libitum from a preweighed vial {±0.001 g) of 20.0% 
sucrose solution for the normal 12-h daylight period. Four 
standard food vials were run concurrently to measure 
evaporative water loss (Chapter 4) . Twenty-four h after the 
initial injection I weighed the bird again (prior to feeding) 
and drew a blood sample from the jugular vein. Following the 
experiment, blood samples were mailed to The Laboratory of 
Biomedical and Environmental Sciences (University of 
California, Los Angeles, CA) for analysis of isotope 
concentrations (Nagy 1983) . Permits for the collection and 
exportation of blood samples were granted by the Ministerio 
de Recursos Naturales Energia y Minas, Direccion General de 
Vida Silvestre, San Jose, Costa Rica (permits 102-88 and 104- 
88) . 

Calculations 

Energy budgets for both methods were calculated to fit 
equation 2.2. For the 24-h period I present I, R, and P in 
units of kJ/d; water flux is presented in mL/d. 

Balance method (BAD . I calculated daily water influx 
from the volume of water consumed as food (corrected for 
evaporation) plus water produced metabolically 

H2OBAL = (Mvi-Mvf-Mevap) { [H2O] + ([SUC] x 0.579)}, (6.1) 



130 



where M^i and M^f are the initial and final feeder masses (g) , 
respectively, Mevap is the mean mass (g) of evaporative loss 
from standard food vials, [HjO] is the water concentration in 
food (0.80 g HsO/g solution), [SUC] is the sucrose 
concentration in food (0.20 g sucrose/g solution), and the 
conversion factor for metabolic water production is 0.579 g 
H20/g sucrose. I assumed that 1 mL of water has a mass of 1 

g- 

I calculated daily energy intake, expenditure, and 
storage with the same gravimetric method described in Chapter 
2. 

Doubly labeled water (DLW) method . Because a pre-trial 
blood sample can severely stress hummingbirds (H. Tiebout, 
pers . obs . ) , I followed the single blood sample procedure of 
Nagy (1983) which was recently validated for small birds 
(Webster and Weathers 1989) . I obtained initial isotope 
activity levels from six additional A. saucerottei not used 
in the feeding trials. I injected these birds exactly as for 
the experimental birds, and drew blood samples (20-40 \Xl) 
after a 40-min equilibration period. I estimated total body 
water (TBW) for these birds using the ^^0 dilution space (Nagy 
1983) and found that the regression for TBW/BM vs. BM was 
not significant. Thus, I used the mean TBW/BM (0.644) to 
estimate TBW from the mass of each experimental bird. As an 
independent check on this TBW/BM value, I also weighed 6 A. 
saucerottei, which were killed and dried to constant mass at 



131 



67 °C. I measured background isotope levels in blood samples 

drawn from three unlabeled A. saucerottei . 

1 calculated daily water output (rnjo) using equation 6 
in Nagy and Costa (1980) and production of CO2 (rco2) using 
equation 2 in Nagy (1980) . These equations do not incorporate 
corrections for fractionation effects. I then recalculated 
values for water flux and CO2 production using the equations 
of Lifson and McClintock (1956) to correct for fractionation. 
For water influx (modified equation 32 of Lifson and 
McClintock 1966) 

r'HjO = 1.033 X KsrN, (6.2) 

where Ksh is the fractional turnover rate for tritium in body 
water (= In (3hi/3h2) /t , where ^Hi and 3h2 are initial and 
final tritium activities in blood samples, and t is time 
elapsed in days), N is average moles total body water [ (Nl + 
N2)/2], and 1.033 is the isotope fractionation correction 
assuming that the physical fractionation factor for tritium 
in liquid water equilibrating with water vapor is 0.84 for Tb 
= 40 °C (Price 1958), and evaporative water loss (EWL) is 20% 
of total water loss (Berger and Hart 1972, 1974) . I then 
converted water influx to mL/d 

K2ODLW = rHaO X 18.0 g/mole x 1 mL/g. (6.3) 



132 



For CO2 production, correction for fractionation effects 
was made by modifying Lifson and McClintock ' s (1966) equation 
35 for tritiated water to yield 

r'C02 = (N/2.08) (K180 - Ksh) - 0.0144(K3hN) (6.4) 

where Kiso = In (l8o;^/i8o2) /t, with 180]_ and ^^02 representing 
initial and final atom% excess in blood samples. I assumed 
that the physical fractionation factors for ^^O in liquid 
water equilibrating with water vapor and gaseous CO2 (0.99 and 
1.04, respectively) were the same as used by Lifson and 
McClintock (1966) and Schoeller, Leitch, and Brown (1986) . 
The value of 0.0144 corrects for fractional evaporation of 
tritiated water in the 20% of total water loss that I assumed 
occurred by evaporation. I converted CO2 production to daily 
energy expenditure 

Rdlw = rco2 (rnoles/d) (22.4 1/mole) (20 . 98 kJ/1) (6.5) 
using the conversion from 1 CO2 to kj for sucrose. 

I calculated daily energy intake from H20dlw with 
metabolic water subtracted 

Idlw = (HzOdlw) ( [SUC]xl6.48)/{ [H20] + ( [SUC]x0.579) } . 

(6.6) 

I calculated daily energy storage from equation (1.2) 

Pdlw = Idlw ~ Rdlw. (6.7) 



133 



Uncorrected estimates (H20dlwia Idlww etc.) were 
calculated from rnjo snd rco?'' corrected estimates (H2ODLW2' 
etc.) were calculated from r'H20 ^^d r'co2- 

As an alternative to the direct DLW estimates, I also 
calculated two indirect estimates of expenditure and intake 

R ' DLW2 "" IdLW2 ~ PbAL (6.8) 

and 

I'dlw2 = Rdlw2 + Pbal. (6.9) 

These estimates serve as a check on the direct DLW estimates 
for R and I, because Pbal is independently derived. In 
addition, equation 6.8 can be derived using singly labeled 
water (HTO) , and hence may be applicable to studies using only 
this method. 



Results 

Total Body Water (TBW) 

The TBW/BM estimates from ^^O dilution space and from 
drying additional birds differed significantly (T = 2.25, P < 
0.05), with the ^^O value (0.644 ±0.018) ca . 4% higher than 
that for dried birds (0.621 ±0.017) . Because changes in BM 
(and hence TBW) were so small over the 24-h trial period (x = 
1%) , calculations of water flux and CO2 production made using 
either of these TBW estimates agreed within ca . 4% (see eq. 2 
in Nagy 1980, and eq. 4 in Nagy and Costa 1980) . I report 



134 



calculations using the ^^O values, because th$y were obtained 
from the same subjects used in the validation trials. 

DLW versus BAL Estimates 

Comparisons among the estimation techniques are 
presented in Table 6-1. For all measures, the DLWi 
(uncorrected for isotopic fractionation) method was 
significantly different from both the BAL and DLW2 (corrected 
for fractionation) methods. The BAL and DLW2 methods did not 
differ significantly, and for all variables these two methods 
had the smallest differences between means of any of the 
methods. The BAL and DLW2 estimates for water flux, I, and R 
agreed within 0.5 to 1.5%. Because estimates of P included 
both positive and negative values, calculations of percent 
difference are not meaningful and are not included in Table 
6-1. For DLW2 and BAL methods, the P means differed by only 
0.59 kJ/d. However, because P was less than 3% of total 
energy intake or expenditure, small errors in estimating I 
and R can result in large relative errors (%) in estimating 
Pdlw- The uncorrected method (DLWi) yielded H2O influx 
estimates 3-5% lower than the other methods, and I and R 
estimates that differed by ca . -5% and +15%, repectively. 
For all estimates (except P) there was also an extremely 
tight one-to-one correspondence between the DLW2 and BAL 
values, as demonstrated by significantly high coefficients of 
correlation (Table 6-1, Fig. 6-1 A-D) . 



135 



In addition, the indirect DLW estimates, I ' dlWj and 
R'dlW2' were significantly correlated with and 
indistinguishable statistically from both the direct DLW2 
estimates, Idlw2 and RdlW2 ' and the corresponding BAL 

estimates (Table 6-1) . 

Discussion 

The significant lack of agreement between the 
uncorrected DLWi and BAL methods suggests that physical 
fractionation of isotopes during evaporative water loss (EWL) 
may be appreciable in the tropical hummingbird species I 
studied. However, when the DLWi values were corrected for 
fractionation using published values, the DLW2 estimates were 
highly similar to gravimetric estimates for all variables. 
Thus, comparisons between studies using either the DLW2 or 
gravimetric method can be made not only in relative terms 
(e.g., %), but also in absolute terms (e.g., kJ/d) . 

The close agreement between DLW2 and BAL estimates for 
water influx alone suggests that water vapor input via lungs 
and skin is not a significant source of error (see Lifson and 
McClintock 1966; Nagy and Costa 1980 for a discussion) . This 
validates tritium as a reliable marker for estimating water 
flux in hummingbirds living in humid tropical environments . 
The significant correlation between both methods for water 
influx, energy intake, and energy expenditure suggests no 
evidence of systematic error in either method and indicates a 



136 



high degree of reliability for each. Thus, despite high 
water flux and metabolic rates, isotope turnover in tropical 
hummingbirds appears not to be affected by dynamic biological 
effects (Lifson and McClintock 1966) . 

This experiment validates the DLW method, corrected for 
physical fractionation, for a tropical hummingbird species in 
a humid environment . Because hummingbirds in temperate zones 
may be less prone to DLW errors due to water vapor input, DLW 
values corrected for EWL fractionation should yield accurate 
estimates of water flux and energy budgets for these birds as 
well. As indicated above, however, when rates of energy 
storage (P) are low relative to I and' R, the DLW method may 
result in appreciable relative error (%), When comparisons 
between low rates of storage are required, gravimetric 
measures of P should be used in conjunction with DLW 
estimates, if possible. 

Finally, the equivalence of Idlw2 vs. Ibal and R ' dlw2 ^^ ■ 
Rbal suggests a simpler and more economical method for 
measuring water flux and energy budgets for certain types of 
cage experiments. When two or more birds share a feeder, but 
individual meal sizes cannot be measured, birds need only be 
injected with tritiated water (HTO) . Assuming a diet with 
precisely known water and energy content, an independent 
determination of TBW/BM, and minimal change in body mass over 
24 h, then H2ODLW2' Idlw2'- R'oLWa^ ^^^ Pdlw2 can be calculated 
from tritium values alone. 



137 



Table 6-1 . Comparisons between doubly labeled water (DLW) 
and gravimetric (BAL) methods for estimating water influx and 
energy budgets for six Amazilia saucerottel . For definitions 
and calculations of terms see text. 





Mean 


± , 


dD 


Versus 


Diff . 


% Diff. 


Water Influx 


(mL/d) 


(mL/d) 


(mL/d) 




H2OBAL 


8.42 


± 


1.68 


H2ODLW1 


0.38 


4.5 


H2ODLW1 


8.04 


± 


1.53 


H2ODLW2 


-0.26 


-3.2 


H2ODLW2 


8.30 


± 


1.58 


H 2 Oral 


-0.12 


-1.4 


Energy Budgets 


(kJ/d) 


(kJ/d) 


(kJ/d) 




Direct Estimates 














I BAL 


30.32 


± 


6.03 


^DLWi 


1.39 


4.6 


^DLWi 


28.93 


± 


5.52 


^DLW2 


-0.95 


-3.3 


^DLW2 


29.88 


± 


5.70 


I BAL 


-0.44 


-1.5 


^BAL 


29.54 


+ 


6.56 


%LWi 


-4.60 


-15.6 


^DLWi 


34.14 


+ 


6.56 


RdLW2 


-4.44 


-13.0 


^DLW2 


29.70 


+ 


5.72 


^BAL 


0.16 


0.5 


^BAL 


.77 


+ 


2.06 


^DLWi 


5.98 


— 


^DLWi 


-5.21 


± 


1.44 


PdLW2 


5.39 


— 


PdLW2 


0.18 


+ 


1.04 


Pbal 


-0.59 


— 


Indirect Estimates 














I 'dlw2 


30.47 


± 


5.98 


I BAL 
IdLW2 


0.15 
0.59 


0.5 

1.9 


R'dlwj 


29.11 


± 


6.20 


^BAL 
^DLW2 


-0.42 
-0.59 


-1.4 
-2.0 



138 



Table 6-1-- continued. 



Water Influx 
H2OBAL 
H2ODLW1 



H2ODLW 



2 



Energy Budgets 
Direct Estimates 



Paired 


t- 


-test 


Correlation 


T-value 




P 


R ■ P 


3.68 




0.014 




12.84 




0.000 




1.28 




0.258 


0.991 0.000 



^BAL 


3.65 


0.015 








^DLWi 


12.84 


0.000 








IdLW2 


1.24 


0.271 


0. 


,991 


0.000 


^BAL 


3.90 


0.011 








^DLWi 


12.86 


0.000 








^DLW2 


0.13 


0.900 


0, 


,898 


0.015 


^BAL 


5.73 


0.002 








^DLWi 


12.86 


0.000 








^DLWj 


0.50 


0.577 


-0, 


.116 


0.827 


Indirect Estimates 












I ' DLW2 


0.13 


0.900 


0, 


.885 


0.019 




0.60 


0.577 


0, 


.915 


0.010 


R'dlw2 


1.24 


0.271 


0, 


.993 


0.000 




0.60 


0.577 





.921 


0.009 



139 



E o 

(0 N^ 




5 6 7 8 9 10 I i 

Water Influx (ml/d) 
(Balance Method) 





"a 
o 

JZ 




•*^ 




0) 




2 




CVJ 


c 


$ 


> 




^ 


>-.' 


o 




c 




UJ 





35- 



25 



y = 


1.488 +0.937X 




./ 


- 


R = 0.991 


/ 


f" 


/ 


, [ 


/ 

IB 


B 



25 



35 



Energy Intake (kJ/d) 
(Balance Method) 



-3 




y- 6.556 +0.783X 




<«-> 
T3 




y = 0.229 -0.058X 




^ 
'*«' 
^-^ 


35- 


R = 0.898 


•■/^ 




'I- 


R=0.114 ■ 




CU T3 








w "o 








if O 








o 








3 x: 




y 




CD -C 








endit 
Met 






» 


orag 
Met 




_^__^^ 


■ 


X $ 


25- 


y^ 




■♦" CM 


0- 


" ~— - 




HI _l 




y^ 




>^ -1 








^, Q 




/■ 




O) Q 




« 








>/ 




!r '^ 








05 

c 

LU 


15- 




c 


0) 

c 

LU 


-h 


■ 

1 : ' 1 1 r 


D 



15 25 35 

Energy Expenditure (kJ/d) 
(Balance Method) 



-4-2024 

Energy Storage (kJ/d) 
(Balance Method) 



Figure 6-1. Correlations between estimates of water influx 
(A) and energy budgets (C-D) for corrected doubly labeled 
water (DLW2) vs. gravimetric (balance) methods for Amazilia 
saucerottei . 



CHAPTER 7 
INTERSPECIFIC COMPETITION FOR LIMITED FOOD 
EFFECTS OF FLOWER DISPERSION 



IntcQ duc t i on 



When nectar is limited, both exploitative and 
interference competition for food will reduce energy 
availability per individual by dividing nectar production 
among two or more consumers. Each bird must then visit more 
flowers, regardless of flower density, resulting in increased 
transit costs per unit nectar consumed. Exploitative 
competition, at any flower density, may further increase 
transit costs above those imposed by reduced energy 
availability alone: as each competitor rushes to harvest 
nectar pre-emptively (Gill 1988) , visit frequency and hence 
transit costs should increase. Alternatively, when flowers 
are sufficiently dense, part or all of a clump may be 
defended. When this happens, interference competition can 
result in energetic costs above transit costs alone, both for 
territory defenders and for intruders, due to increased 
flight time associated with chases and other aggressive 
behavior (Chapter 4) . 



140 



141 



Thus, variation in the ecological factors of nectar 
production, flower dispersion, flower defensibility, and the 
number and types of competitors can change foraging behavior 
and energetics of hummingbirds. My final experiment examined 
the consequences of interspecific competition for limited 
food resources at each of two patterns of feeder 
distribution: clumped versus dispersed. I chose the food 
level and competition type to provide the most parsimonious 
test of the effects due to species and defensibility. 

Based on my previous experiments and on field 
observations, I expected that the consequences of food 
competition would be asymmetrical for Amazilla and 
Chlorostilbon . In my second experiment (Chapter 4), with an 
easily defensible ad libitum feeder, I found the strongest 
and most asymmetrical energetic responses to competition were 
for heterospecif ic rather than conspecific pairs. I expected 
more marked energetic effects in response to interspecific 
competition for limited food, because Chlorostilbon would 
likely experience a greater discrepancy between energy intake 
and expenditure than with unlimited food. In addition, in my 
third experiment (Chapter 5) , under conditions of no 
competition, I found that feeder dispersion had important and 
different energetic effects on the territorialist and the 
trapliner, particularly when food was limited. I expected 
that these differences might also be enhanced under 
interspecific competition, especially when clumped feeders 
were defensible by Amazilia . Finally, Feinsinger (1976) 



142 



found that most interspecific aggressive encounters between 
Amazilia and Chlorostilbon occurred when flowers were dense. 
Under these conditions, he found that Amazilia had a 
relatively large competitive effect on the local population 
of the low-reward trapliner, whereas Chlorostilbon had 
relatively little competitive effect on the territorialist 

(see Chapter 4) . I expected that similar asyminetries might 
be observed in the energetic responses of individual Amazilia 
and Chlorostilbon sharing clumped resources in a cage. 

I tested heterospecif ic pairs in a flight cage at two 
feeder dispersion treatments. In the clumped treatment five 
feeders were easily defensible from a single perch. In the 
dispersed treatment only one or two feeders were defensible 
at a time . When feeders were clumped, I expected the 
territorialist, Amazilia, to do better energetically than the 
trapliner, Chlorostilbon, and to do better than Amazilia at 
dispersed feeders. In contrast, when feeders were dispersed 
I expected Chlorostilbon to fare better than Amazilia, and 
better than Chlorostilbon at clumped feeders. In addition to 
testing these general predictions, I measured visit diversity 

(H') and feeder overlap between cagemates (PSI) to determine 
if spatio-temporal patterns of resource exploitation varied 
by species, by feeder dispersion, or over time. 



143 
Materi a ls a nd M et h ods 
Experimental Flight Cage 

I tested heterospecif ic pairs in the flight tube 
previously described (Fig. 5-1), with the following 
modifications (Fig. 7-1) . Each tube segment I-IV had a 
single perch large enough to accommodate easily two birds. 
Segment V had two perches, so as to reduce aggression during 
the initial 30-min period, when birds were held behind the 
retention curtain. 

In the dispersed feeder treatment, each tube segment had 
a single feeder at the end opposite each perch. In this 
configuration each feeder was at least 4 m from the nearest 
perch, and only one feeder could be seen from any given 
perch. The total distance between all feeders was at least 
16 m. For these reasons, it was impossible for a bird to 
defend more than one feeder from a given perch, or two 
feeders while in flight, unless the bird followed its 
cagemate from feeder to feeder (never observed) . In 
addition, some feeders, even when visible from a given perch, 
could be approached from the "blind side" (e.g., approaching 
the feeder in segment II from segment III) . Thus, a defender 
on perch II would not be aware of an intruder until the 
latter was already at the feeder. This further reduced 
def ensibility of the feeders in the dispersed feeder 
treatment . 



144 



In the clumped feeders treatment, the sape five feeders 
were arranged in a diamond pattern (50 x 50 cm) located at 

the midpoint of segment III. In this configuration the 
feeders could be seen only from perch III, from which they 
were only 2 m. Thus, all five feeders were easily defensible 
by a bird on perch III or in flight anywhere in segment III. 
Furthermore, unlike the dispersed configuration, an intruder 
could not approach feeders without traversing 2 m of open 
airspace in segment III. Thus, intruders could be easily 
spotted and repelled before gaining access to the feeders. 

Protocol 

I mist-netted hummingbirds during May and June 1988 and 
housed them as described in Chapter 2 . Three or 4 d before 
its first trial, each subject was allowed to learn its way 
around the flight tube. Following a protocol similar to that 
in Chapter 5, but using conspecific pairs instead of 
individuals, birds visited feeders in the dispersed 
arrangement (see below) for 2 h. Amazllia and Chlorostilbon 
did not differ in the time it took to discover and visit each 
of the five feeders. Two days before a trial, a subject of 
each species was moved to an individual 1 m^ holding cage and 
maintained on 20% sucrose solution and Drosophila, ad 
libitum. The day before a trial, each bird was moved to the 
flight-tube room and housed in an individual 0.75 m-^ holding 



145 



cage. This allowed birds to adjust to the experimental 
feeders and room conditions . 

Between 0445 and 0545 on the day of a trial, birds were 
weighed (±0.001 g) and injected intramuscularly with the 
doubly labeled isotope mixture (Chapter 6) . Amazilia 
received 50.0 )1L H2O containing 97 atom% ^^0 and 38 |lCi 
tritium; Chlorostilbon received 30.0 |J,L H2O containing 97 
atom% i^O and 23 jlCi tritium. Each bird was then returned to 
its holding cage and allowed to feed ad libitum from a pre- 
weighed vial until 0645. Both birds were then moved to the 
flight tube (Fig. 7-1), where they were kept behind the 
retention curtain without food for 30 min . This allowed 
birds to adjust to the flight tube and provided sufficient 
time for birds to empty the contents of their crops 
(Hainsworth and Wolf 1972a, see Chapter 3) . This ensured 
that birds had calmed down from the cage transfer and were 
ready to feed. At 0715 the first delivery of sucrose 
solution was made to the experimental feeders, and the 
curtain was opened to begin the trial. 

Sucrose solution was delivered to the five feeders 
simultaneously via microtubing, as described in Chapter 5, at 
15-min intervals until 1600, for a total of 7800 |IL . Thus, 
on average, each feeder received 43.3 p,L per delivery. This 
rate was sufficient to keep birds active while keeping them 
below satiation levels. I recorded behavioral data on focal 
birds, alternating species, during 12 20-min observation 
periods per day, beginning at 0830, 0900, 0930, 1000, 1030, 



146 



1100, 1300, 1330, 1400, 1430, 1500, and 1530. At 1630 I 
returned each bird to its own holding cage, where it fed ad 
libitum on 20% sucrose solution from a pre-weighed vial until 
lights were turned off at 1745. The following morning, 24-h 
(±10 min) after isotope injection each bird was weighed 
(±0.001 g) , and a blood sample was drawn (20-40 )ll) and 
stored under refrigeration in flame-sealed microhematocrit 
capillary tubes . The two body mass measurements for each 
bird were used to calculate its mean body mass for the trial. 
At the end of the experiment samples were mailed to the 
Laboratory of Biom.edical and Environmental Sciences, 
University of California, Los A-ngeles, for isotope analysis 
(Nagy 1983) . 

Experimental Design and Analyses 

I ran 10 individuals of each species for each feeder 
treatment (clumped and dispersed) . I ran each individual at 
both treatm.ents, if possible, with treatment order randomly 
assigned. Due to the death of several birds, some 
individuals received only one treatment and were replaced by 
new birds for the second. Thus, a total of 11 Amazilia and 
15 Chlorostilbon were tested (mean body masses: 4.04 ±0.38 
and 2.10 ±0.16 g, respectively) . 

During the observation period for a given focal bird, I 
measured flight time and recorded visits (Chapters 2 and 5) 
to each of the five feeders by number (see Fig. 7-1) . 



147 



Agonistic acts initiated and received (Chapter 2) within 1 m 

of feeders were scored as feeder defenses and those further 

from feeders as perch defenses, although the latter also 

included a small proportion of miscellaneous chases and body 

contact not clearly associated with either feeders or 

perches. Perch and feeder defenses in a given time period 

were also summed for total agonistic acts initiated or 

received (per individual) . Because agonistic acts 

necessarily involved both birds, this behavior was scored for 

both the focal bird and its cagemate simultaneously during' 

all 12 observation periods. Flight time and visit behavior, 

however, could be monitored only for a single bird (focal 

bird) during a given observation period and were recorded for 

only 6 observation periods per individual for each trial. 

For each of six observation periods I also calculated 

two measures of the pattern of feeder visitation: visit 

diversity and feeder overlap. Visit diversity was calculated 

using the Shannon-Wiener diversity index 

5 
H' = - E(pi Xln(pi) ), (7.i; 

i=l 

.where H' is diversity and Pi is the proportion of visits by a 
focal bird to feeder i. Feeder overlap was calculated using 
the Proportional Similarity Index (PSI) 



148 



PSI = 1 - 0.5 Slpi _ qil, (7.2; 

■ i=l 



where pi and q± are the proportions of visits to each feeder i 
(1 through 5) made by each of two cagemates, p and q. PSI 
sums the proportions of overlap in feeder visits, with index 
values ranging from (no overlap) to 1.0 (complete overlap). 

Energy budgets were calculated from DLW analyses as 
described in Chapter 5 and presented as mass-specific rates 
by dividing time rates by each bird's mean body mass (see 
above) . In addition, energy intake for each bird was 
calculated for the periods (a) before entering and (b) after 
leaving the flight tube, and (c) during the flight tube trial 

ItRIAL = ItOTAL - (^PRETRIAL + IpOSTTRIAl) ^ (7.3) 

■where Itrial is energy intake while in the flight tube, Itotal 
is total intake (calculated using the DLW method) , and 
^PRETRIAL and IposTTRiAL are intake before and after the flight 
tube trial, respectively (calculated gravimetrically from the 
pre-weighed food vials, the BAL method from Chapter 6) . I 
also calculated a second estimate of energy storage using the 
BAL method for comparing with the DLW estimates. Terms of 
the energy budget are presented as J ■ g~l ■ h~l for trials (= 
time in the flight tube, intake only) and kj-g-l-d~^ for the 
entire 24-h period (ca. 0515-0515) . 



149 



I tested dependent variables using a 2-way ANOVA, with 
species as the grouping variable and feeder treatment as the 
trial factor. Behavioral data were also tested for time-of- 
day effects by adding observation period as repeated measures 
to the ANOVA model. Post hoc group comparisons were made 
using independent t-tests between species or feeder 
treatments, and paired t-tests between periods for a given 
species and feeder treatment . 

Results 

Behavior 

Species, feeder dispersion, and time of day had 
significant effects on the frequency of feeder visits 
(Fig. 7-2) . In general, both species decreased visit rate 
later in the day (period effect: F = 11.96, P < 0.001; Fig. 
7-2) . Visit rates in the morning were high, with a mean for 
periods 1-3 of 159.4 visits/h. Rates dropped substantially 
in the afternoon, decreasing 66% to a mean of 54.0 visits/h. 
Even these lower visit rates are high compared to solitary 
birds feeding ad lihitum (see Chapter 4) . Relatively high 
visit rates, coupled with the observation that all feeders 
were usually drained within a few minutes of each delivery, 
indicate that at least one cagemate was likely to remain 
unsatiated during each trial. Species differed in their 
responses, both in magnitude and pattern over time. During 
both feeder treatments, Amazilia made significantly more 



150 



visits than Chlorostilbon (135.1 ±17.1 vs. 78.4 ±135.7 
visits/h, respectively; F = 4.02, P < 0.05; Figs. 7-2 and 7- 
3) . This effect was most pronounced in the morning (206.4 
vs. 112.6 visits/h for periods 1-3, Amazilia and 
Chlorostilbon, respectively) , with only a relatively small 
interspecific difference during the afternoon (63.8 vs. 44.2 
visits/h for periods 4-6, Amazilia and Chlorostilbon, 
respectively) . 

In general, birds visited more frequently when feeders 
were dispersed than when clumped (119.2 ±145.8 vs. 94.2 
±182.7 visits/h, respectively; F = 5.34, P < 0.05; Fig. 7-2 
and 7-3) . When feeders were dispersed, both species had 
slight early morning peaks in visitation (periods 1 or 2) 
followed by a reduction in visit rate over the rest of the 
day. Amazilia reduced visit rate abruptly between periods 3 
and 4, whereas Chlorostilbon reduced visit rate more 
gradually throughout the day. In contrast, when feeders were 
clumped, birds peaked in visitation later (period 3) , with 
the same substantial decline during the afternoon (period x 

feeder treatment interaction: F = 2.30, P < 0.05) . There was 

a trend for species to respond differently over time (species 
X period interaction: F = 2.73, P = 0.1), which v-ras due 

primarily to the unusual pattern exhibited by Chlorostilbon 
visiting clumped feeders. Contrary to the typical pattern, 
Chlorostilbon made very few visits early and late in the day, 
but had a marked peak during period 4. . , 



151 



Agonistic behavior showed marked effects due to species, 
feeder treatment, and time of day (Fig. 7-4 and 7-5) . All 
measures of aggression initiated (perch defense, feeder 
defense, and total agonistic acts) were higher for both 
species when feeders were clumped (treatment effects: F = 
6.68 to 29.78, P = 0.001 to 0.014; both species combined). 
These measures were higher in both feeder treatments for 
Amazilia compared to Chlorostilbon (species effects: 
F = 262.2 to 686.8, P < 0.001; both treatments combined). 
Extreme mean daily rates of total agonistic acts initiated by 
individuals ranged from 0.50 acts/h for Chlorostilbon with 
dispersed feeders to 188.25 acts/h for Amazilia with clumped 
feeders. Amazilia was capable of short periods of extremely 
intense aggression when defending the clumped feeders. 
During one 20-min observation period, one Amazilia defended 
the feeder 117 times and the perch 4 timies, for a total rate 
of 363 agonistic acts/h (or ca . 1 act/10 s) . In contrast, 
Chlorostilbon only very rarely initiated aggressive acts, 
with a high of 5 feeder and 5 perch defenses in one 20-min 
period (or 1 act/2 min) when feeders were clumped. 

Aggressiveness by Amazilia changed over the course of 
the foraging day, with the pattern dependent upon feeder 
treatment (Fig. 7-5) . When feeders were dispersed, total 
agonistic acts initiated by Amazilia declined monotonically 
throughout the day. The early morning high of ca . 18 acts/h 
dropped to a late afternoon low of under 4 acts/h. In 
contrast, when feeders were clumped Amazilia started and 



152 



ended with high rates of aggression (ca. 40 and 52 acts/h, 
respectively) with a significant depression occurring during 
midday (periods 6-7: 26 acts/h) . 

Birds averaged from 21-27% of the day in flight, without 
significant effects due to either species or feeder treatment 
(Fig. 7-6) . Flight time decreased significantly over the 
foraging day (period effect: F = 7.48, P = 0.05) and at 
different rates for each feeder treatment (period X feeder 

treatment interaction: F = 2.22, P < 0.001; Fig. 7-7) . When 
feeders were dispersed flight time declined monotonically, 
whereas when feeders were clumped birds maintained nearly 
constant flight times throughout the day. 

There was a trend for birds to decrease overlap (PSI) in 
their use of feeders as the day progressed (period effect: 
F = 2.04, P = 0.08; Fig. 7-8). Birds significantly reduced 
overlap by ca . 36% from the first period high (54.7 ±23.8%) 
to the final period low (34.9 ±23.5%; T = 3.04, P < 0.01) . 
Feeder treatment had no significant effect on overlap, 
although mean overlap for clumped feeders was slightly more 
(up 15%) than for dispersed feeders (48.5 ±24.3% vs. 42.2 
±29.7%, respectively; F = 8.80, P > 0.1). Figure 7-9 
illustrates the reduction in overlap over time, as shown by a 
typical pair of birds sharing dispersed feeders. 

The diversity of feeders visited (H') showed significant 
effects of both time of day and species (Figs. 7-10 and 
7-11) . For every period Amazilia visited a greater diversity 
of feeders than did Chlorostilbon (1.13 ±0.44 vs. 0.87 ±0.57, 



153 



respectively; P < 0.005) . In general, both species exhibited 
a decrease in visit diversity over time (period effect: F = 
3.71, P < 0.005), with Chlorostilbon showing a more marked 
decline. Both species started the day with equally high 
visit diversity (1.1 - 1.2), but finished the day widely 
divergent (1.11 ±0.42 vs. 0.59 ±0.47, Amazilia and 
Chlorostilbon, respectively; T = 3.73, P < 0.001). Visit 
diversity exhibited no main effect of feeder treatment (1.05 
±0.44 vs. 0.95 ±0.59, dispersed vs. clumped, respectively), 
but there was a significant interaction between period and 
feeder treatment (F = 4.15, P = 0.001) . For the first three 
periods, during which 74.7% of all visits occurred, diversity 
increased for clumped feeders and decreased for dispersed 
feeders . The pattern for the last three periods was 
variable . 

Energetics 

Rates of energy intake, both during trials (= time in- 
flight tube) and for 24 h, exhibited similar patterns (Fig. 
7-12) . Both species had equivalent rates when feeders were 
clumped (552.0 ±128.4 J-g-i-h'l and 7.26 ±1.31 kJ-g-l-d"!, 
trials and 24-h, respectively) . When feeders were dispersed, 
however, Amazilia had slightly reduced intake, whereas 
Chlorostilbon had increased intake (species x feeder 

treatment interaction: F = 3.79, P = 0.059). This resulted 
in significant interspecific differences, with Chlorostilbon 



154 



having intake rates 2 6.9% higher than Amazilia during trials 
with dispersed feeders (617.4 ±148.5 vs. 486.6 ±57.0 
kJ-g-l-h-l, respectively, T = 2.60, P < 0.05; for dispersed 
feeders only) . Differences in visit rate were reflected in 
the rates of energy intake per feeder visit (Fig. 7-13) . 
Birds obtained food at a significantly higher rate per visit 
when feeders were clumped (increase of 113% over dispersed 
feeders, F = 5.92, P < 0.05). Neither feeder treatment nor 
species had an effect on food consumption following trials 
before roosting (P > 0.1) . 

Rates of energy expenditure over 24 h differed greatly 
between species for both feeder treatments (Fig. 7-14). 
Chlorostilbon had consistently higher expenditure rates, 
averaging 26% more than Amazilia (species effect: F = 22.25, 
P < 0.001) . The differences between species showed a trend 
of increasing when feeders were dispersed compared to clumped 
(species x feeder treatment interaction: F = 2.52, P = 0.1). 

Both measures of energy storage agreed in direction but 
not in magnitude of treatment effects (Fig. 7-15) . The DLW 
and BAL estimates indicated that Amazilia had greater 
storage rates than Chlorostilbon (DLW method: F = 18.55, P 
< 0.001; BAL method: F = 2.83, P = 0.1), with no effects of 
feeder treatment. The DLW method indicated that Amazilia 
maintained slightly positive energy storage over 24 h (0.121 
±0.569 kJ-g~l-d~^), whereas Chlorostilbon lost energy stores 
(-1.022 ±1.022 kJ-g-l-d-1). In contrast, the BAL method 
indicated that both species maintained slightly positive 



155 



energy storage (0.545 ±1.123 kJ-g i-d~^, both species 
combined) . 

D i sciss ion 

Behavioral and Energetic Responses 

During both feeder treatments, Amazilia and 
Chlorostilbon exhibited similar patterns of visitation. For 
both species combined, birds reduced visit rate, visit 
diversity, feeder overlap, and flight time during a trial. 
Because the rate of food delivery remained constant, birds 
foraged with increasing efficiency throughout the day by 
reducing visits and flight time and by concentrating on a 
smaller subset of feeders that were likely not visited by a 
cagemate . In the trials with dispersed feeders, aggression 
levels also dropped monotonically throughout the day. This 
suggests that when food sources are dispersed, these 
hummingbirds may use trial and error to sort out patterns of 
utilization that reduce overlap and conflict while minimizing 
expenditures. Minimizing expenditures can increase foraging 
efficiency, and hence profit, as individuals divide the 
resource base into a number of nearly non-overlapping, 
relatively fixed foraging routes. This division of resources 
could occur over a short period of time simply as a result of 
each participant striving to maximize its own foraging 
efficiency . 



156 



Changes in feeder density had a major impact on bird 
behavior, which varied with species. The midday peak in 
visitation exhibited by Chlorostilbon when feeders were 
clumped contrasts sharply with the pattern shown by the other 
three groups (Fig. 7-2), and is the mirror image of the 
frequency of agonistic acts initiated by Amazilia during the 
trials with clumped feeders (Fig. 7-5: solid symbols). Thus, 
Chlorostilbon made more feeder visits during periods when 
Amazilia was less aggressive, and fewer when Amazilia was 
most defensive of clumped feeders. In contrast, when feeders 
were dispersed, there was no apparent effect of Amazilia^ s 
aggressiveness on visit rate by Chlorostilbon . 

Despite strong interspecific differences in both flight- 
dependent behavior patterns, visit rate and aggressiveness, 
Amazilia and Chlorostilbon did not differ in flight time. 
The two species did, however, differ markedly in energy 
expenditure, with consistently higher rates for the trapliner 
Chlorostilbon. This probably reflects non-flight metabolic 
rates, such as higher resting metabolic rates for 
Chlorostilbon (see Chapter 5 for allometric relationships) or 
possible use of nighttime torpor by Amazilia (not observed) . 
The trend for Chlorostilbon to have slightly elevated 
expenditures when feeders were dispersed does coincide with a 
small increase in flight time (up 28% compared to clumped 
feeders, P > 0.1), which may have contributed to its higher 
expenditures . 



157 



Despite the tenacious defense of clumped feeders by 
Amazilia, intruding Chlorostilbcn were able to maintain rates 
of food intake equivalent to territory defenders. 
Chlorostilbon apparently had the greatest intake rates during 
the midday periods, when it was able to make the most feeder 
visits. During midday periods, Chlorostilbon also visited a 
greater diversity of feeders and overlapped the miost with 
Amazilia in feeder utilization. In contrast, when visit 
rates were lowest for Chlorostilbon, so were diversity and 
overlap. 

When feeders were dispersed, Chlorostilbon achieved 
intake rates more than one-fourth greater than Amazilia . 
This success of Chlorostilbon relative to Amazilia, and 
relative to Chlorostilbon at clumped feeders, was probably 
due largely to reduced effectiveness by the territorialist in 
excluding the trapliner from feeders when they were widely 
dispersed. This increased the likelihood that the trapliner 
would visit a feeder that had recently received a food 
delivery before Amazilia . 

Both species had slightly higher visit diversity when 
feeders were dispersed (up ca . 10%), accompanied by a slight 
decrease in their overlap in feeder use (down ca . 18%) . 
Thus, when feeders were clumped, both species may have 
concentrated visits on a small but simiilar subset of feeders 
during each period. In contrast, when feeders were 
dispersed, each species visited a greater variety of feeders 
but with less overlap. Because Chlorostilbon maintained 



158 



higher intake rates than Amazilla when feeders were 
dispersed, despite the observation that Chlorostilbon had 
both lower visit rates and diversity than Amazilla, 
Chlorostilbon may have timed visits better to coincide with 
food deliveries than did Amazilla . Although the trapliner 
appeared to be a successful competitor for food, it paid a 
much higher cost in terms of energy expenditure than did the 
territorialist and thus experienced reduced energy storage 
relative to Amazilla for both feeder treatments. 

Relative Energetic Success of Each Foraging Mode 

As expected, the territorialist Amazilla obtained a 
greater intake rate when it could successfully defend feeders 
than when it could not. Contrary to expectation, however, 
Amazilla did not obtain food at a significantly higher rate 
than its trapliner cagemate when feeders were clumped. 
Combined with significantly lower rates of expenditure for 
Amazilla, though, the territorialist succeeded in maintaining 
positive 24-h energy storage rates when feeders were clumped. 
When feeders were dispersed, the territorialist suffered 
significantly lower intake rates than the trapliner, as 
expected. Despite this reduced intake, Amazilla also reduced 
energy expenditure by an equivalent amount and thus 
maintained positive 24-h energy storage. Because Amazilla 
did not reduce flight time when feeders were dispersed, this 
energy savings probably resulted either from reduced resting 



159 



expenditures (nighttime torpor or DHC, see Chapter 5) or 
because flying while foraging is energetically less expensive 
than flying during aggressive encounters (see Chapter 4) . 
Flight during agonistic interactions could be more costly 
than foraging flight if the former (a) consisted of a higher 
ratio of hovering to forward flight (Hainsworth et al . 1981), 
or (b) consisted of a greater proportion of flight at 
suboptimal speeds (Pennycuick 1968, Gill 1985), including 
more stopping, starting, and rapid acceleration. The use of 
torpor or DHC could indicate that Amazilia had inadequate 
daytime energy storage during the trials with dispersed 
feeders. Thus, in terms of daytime energy management, the 
territorialist may have fared somewhat poorly compared to the 
trapliner when feeders were dispersed. 

The trapliner Chlorostilbon, as expected, had higher 

intake rates at dispersed feeders compared to clumped, and 
higher intake rates than Amazilia at dispersed feeders. 
However, with dispersed feeders, Chlorostilbon increased 
flight time and energy expenditure, relative to clumped 
feeders, by an amount equivalent to its increase in intake. 
Furthermore, at both feeder treatments Chlorostilbon had 
higher rates of expenditure than intake, resulting in lower 
energy storage rates for the trapliner compared to the 
territorialist . 



160 



CLUMPED FEEDERS TREATMENT 



SEGMEr^S: 

I II 



III 



4 m 




PERCHES 



RETENTION 
CURTAIN 

CLUMPED 
FEEDERS 



DISPERSED FEEDERS TREATMENT 



^X" 


# 






1 — 1 1 — ^_j 

^ ^ ^ A^ ^ ^ ^ 

V 








* 








^ 




/ 




V 




/ 




^. 


B 


/ 


t 




V 


1 


Y^''- 


* 
,,Xr i 


1 1 




# 


1 


© 




@ 


© 




M 



DISPERSED 
FEEDERS 



Figure 7-1. Experimental flight tube. Perches numbered 1-5 
following segment numbers. (A) Clumped feeders treatment: 
feeders numbered 1-5, left to right starting at top. (B) 
Dispersed feeders treatment: feeders numbered 1-5 by segment 
number. See Fig. 5-1 for details of observation procedure 
and Chapter 5 for construction. 



161 



m DC 

^ o 

LU ,v- 

Q 

UJ 

\n 

LL 



111 



■ Amazilia DISPERSED 
D Amazilia CLUMPED 

Chlorostilbon DISPERSED 
Chlorostilbon CLUMPED 




PERIOD 



Figure 7-2. Feeder visits for each observation period. Each 
column is for 10 individuals (mean ±SE) . 



«■ t^^c;^'* •'< 'TH'' — ; 



162 



(fi 
t 

CO ir 

X 



LU 
Q 
LU 
LU 



cr 

UJ 

a. 



200 1 



150- 



100- 



50- 



Amazilia 
Chlorostilbon 








DISPERSED 



CLUMPED 



FEEDER TREATMENT 



Figure 7-3. Feeder visits for all observation periods 
combined. Each point is for 10 individuals (mean ±SE) . 
text for description of feeder treatments. 



See 



163 



DC 

CO o 

Lli X 
CO 

Lu az 

U- UJ 

UJ CL 
O 

T- Q 

O ^ 

ai ^ 

QL I- 



on - 


— o— 

A 


Amazilia 
Chlorostilbon 




15: 

■ 


A 


J^ 


-^ 


10^ 


c 








5i 










0: 

• 
; 




A 




A 


M 




A 


-^ 


DISPERSED 




CLUMPED 



CO z> 

m o 

^ X 

LU 

LLI O 

Q LU 

LU H- 

LU < 



70: 
50: 
so- 
lo 

-10 




DISPERSED 



CLUMPED 



Figure 7-4 . Aggressive behavior for all observation periods 
combined. (A) Perch defenses, and (B) feeder defenses. 
Symbols and feeder treatments as for Fig. 7-3. 



164 







70 


(n 






o 




60 


< < 






-J 




50 


O N 


^_^ 






Q 


40 


Z < 


Jd 




o 


k_ 


30 


ss 


a. 


20 


<l 




10 


H ^ 






o 






h- 




U 

-10 



O DISPERSED FEEDERS 
• CLUMPED FEEDERS 



y = 44.80 - 4.638X + 0.167x^2 + . 00232x^^3 
R = 0.951 




17.53 - 1.165X R = 0.867 



01 2345678 

PERIOD 



-I i 1 1 

9 10 1112 



Figure 7-5. Aggressive encounters won by Amazilia as a 
function of time. Each point is for 10 individuals (mean 
-SE) . Both regression lines significant (P < 0.05) . See 
text for treatment descriptions. 



165 



o 
z 

cr 

UJ 
H > 

Z O 

UJ X 

Q. ^ 

^ a ^ 

"^ 1 ^ 
1 

•- o 

z 

> 

-J 
u. 



35 1 
30 

25: 

20- 



15 



Amazilia 
"A — Chlorostilbdn 




DISPERSED CLUMPED 

FEEDER DISPERSION 



Figure 7-6. Flight time for all observation periods 
combined. Symbols and treatments as for Fig. 7-3. 



o 
z 

E 

m 

> 

H O 

Z I 

m 

UJ < ■-- 
= O 



> 
-J 
u_ 



40- 



30: i 



20- 



10- 



DISPERSED FEEDERS 
CLUMPED FEEDERS 




3 4 5 

PERIOD 

Figure 7-7. Flight time for each observation period, 
text for treatment descriptions. 



See 



156 



0.60-1 




0.30 



4 



6 



a. 
< 

cc 

LU 

> 

111 ^ 

Q 
lU 

m 

LL 



0.25 



DISPERSED FEEDERS 
CLUMPED FEEDERS 




2 3 4 

PERIOD 



Figure 7-8. Overlap (PSI) between Amazilia and Chlorostilbon 
in use of feeders by observation period. See equation 7.2 for 
calculation of PSI. (A) Overlap for all treatments combined. 
(B) Overlap by feeder treatment. 



167 



■| Amazilia 

I I Chlorostilbon 



PERIOD 



o 

z 

UJ 

o 

GC 
LL 

H 




1.0 
0.8 
0.6 
0.4 
0.2 
0.0 



PERIOD 4 (0) 




1 



PERIOD 



(.49) 




1.0 

0.81 

0.6 

0.41 

0.2 

0.0 



RIOD 6 (0) 




FEEDER NUMBER 



Figure 7-9. Changes in visit frequency per feeder over time 
(period 1 to 6) for one pair {Amazilia 55 and Chlorostilbon 
55) sharing dispersed feeders. Feeder numbers follow flight 
tube segment numbers (Fig. 7-1) . Feeder overlap (PSI, Eqn . 
7-2) follows period number in parentheses. 



168 



t 1 .20 i 

C/) 

S 1.10-i 

> ^ 

Q X 1.00: 



H 0.90 d 

> 0.80 




...^ 



. " ■: 



T 



i'A 



. ^ 



T 




(/) 

LU ^ 

> X 



> 



Amazilia 
Chlorostilbon 




CO 

a: 

LU 

> ^ 

> 




2 3 4 

PERIOD 



Figure 7-10. Visit diversity by observation period. (A) All 
treatments combined, (B) by species, and (C) by feeder 
treatment (solid bars = dispersed, hatched bars = clumped 
feeders) . 



169 



>- 

55 
tr 

> ^ 



> 



1.3 

1.2 

1.1 

1.0 

0.9-1 

0.8 



0.7 



Amazilia 
Chlorostilbon 




DISPERSED CLUMPED 

FEEDER TREATMENT 



Figure 7-11. Visit diversity for all observation periods 
combined. Symbols and treatments as for Fig. 7-3. 






170 



III UJ 



< 

I- 



>- 
O 
QC 
LU 

Z 
yj 



3 
I- 

H 




650- 
600- 
550- 
500- 
450- 
400 



AmazHia 
Chlorostilbon 




DISPERSED 



CLUMPED 



LU CO 
^ DC 



< 
Z 

> 

o 
a: 

UJ 

z 

LU 



o 

X 
CM 

OC 
LU 

> 

o 



8.5- 
8.0- 
7.5- 
7.0- 
6.5- 



6.0 




DISPERSED CLUMPED 

FEEDER TREATMENT 



Figure 7-12. Energy intake. Symbols and treatments as for 
Fig. 7-3. (A) Hourly intake (J'g~l-h-l) while birds were in 

flight tube only. (B) Intake per day (kj • g~l • d""i) (includes 
intake before and after birds were in flight tube, see text 
for protocol) . 



171 



m 


(/) 


< 


> 


1- 


cc 




LXJ 




O 


> 


LU 


o 


LU 


q: 


IL. 


LU 




z 


CC 


LU 


LU 




Q. 



40 n 



30- 



20- 



10 







-Q — Amazilia 

-A — Chlorostilbon 




DISPERSED 



CLUMPED 



Figure 7-13. Energy intake rate per feeder visit 
( J • g"-"- • h~i • visit"^) for all observation periods combined, 
Symbols and treatments as for Fig. 7-3. 



LU 




CC 




=5 


0) 


H 


CC 


Q 


3 


Z 


o 


LU 


X 


Q. 




X 


^ 


LU 


CM 


>- 


tr 


(D 


LU 


CC 


> 


LU 


o 



LU 



9.0- 



8.0 



7.0 



6.0 




DISPERSED CLUMPED 

FEEDER TREATMENT 



Figure 7-14. Energy expenditure per day (kJ-g ^-d 1) 
Symbols and treatments as for Fig. 7-3. 



172 



LU 
O 

< 

cc 

o 

H 
0) 



Q 
O 

X 
H 



tr -J 

LU 



0.5i 



0.0- 



-0.5- 



■1.0- 



-1.5 



-O — Amazilia 

■A — Chlorostilbon 





DISPERSED 



CLUMPED 



lU 
O 
< 

O 

H 



(/} 111 



i- 1.00-j 

O 
X 



> 
O 
CC 

m 

z 
m 



< 



0.50- 



0.00 



-0.50 




DISPERSED CLUMPED 

FEEDER TREATIVIENT 



Figure 7-15. Energy storage per day (kJ • g~l • d~^ ) . Symbols 
and treatments as for Fig. 7-3. (A) Calculated using the. 
doubly labeled water method (DLW) . (B) Calculated using the 
gravimetric or balance method (BAL) . See text for formiulae. 



CHAPTER 8 
GENERAL DISCUSSION AND CONCLUSIONS 



In this chapter I briefly summarize the results of 
Chapters 3,4,5, and 7, and relate them to' natural communities. 
First, I summiarize the most important conclusions from each 
experiment. Second, I present a few general themes that 
appear throughout the experiments and that may be relevant to 
birds in the field. Finally, I evaluate the use of the 
mechanistic experimental approach and discuss the possible 
role of energetic differences among foraging modes in shaping 
guild composition. 



Implications of Results to Amazilia and Chlorostilbon 

in the Wild 



Caged experiments are by necessity simplistic 
representations of natural conditions, and their results must 
therefore be generalized with caution. The confidence with 
which captive studies can be generalized to wild birds 
depends largely upon the fidelity with which experimental 
manipulations simulated natural factors and the degree to 
which captives behaved as they would in the wild. The former 
I address for each experiment separately in its respective 
section below. The latter I discuss immediately below. 



173 



174 



Two lines of evidence suggest that captive birds 
responded as naturally as possible for confined animals. 
First, birds appeared to adjust to captivity extremely 
rapidly. Wild-caught hummingbirds usually learned to visit 
artificial feeders within a few hours. As soon as new 
captives learned to feed themselves, they abandoned efforts 
to escape (or behavior that appeared as such) . Captive birds 
typically maintained body mass and apparently excellent 
health for many months indoors, and frequently visited 
feeders near the lab after their release. Second, captive 
birds exhibited a wide range of behavior typical of free- 
ranging birds, and I observed no unusual patterns among 
captives. In the communal aviary birds fed from and defended 
the numerous artificial feeders. Captives of both species 
readily hawked for insects, bathed in fresh water, called and 
displayed from perches and in flight, and "explored" their 
surroundings. After less than a week in captivity, birds 
seemed relatively indifferent to my presence and exhibited 
all of the above behavior while observed quietly from a:s 
close as 1 m. Although qualitatively the birds appeared to 
behave normally, quantitative differences in behavior between 
caged and wild birds are possible. 

Learning probably played an important role in shaping 
birds' behavior in the cage experiments. First, effects of 
learning that had occurred prior to capture may have 
predisposed individuals to certain types of responses. 
Though adaptive in the wild, some of these responses may have 



175 



appeared inappropriate to the ecological conditions simulated 
in the experiments. Second, birds may have learned novel 
behavior patterns while caged, both in the communal aviary 
and during the course of actual trials . I attempted to 
control for learning effects through random assignment of 
treatments (when a given subject was used in a repeated- 
measures design) and by standardizing somewhat the time spent 
between trials in the aviary and holding cages. Finally, 
subjects were not used in more than one experiment (except 
for the DLW validation. Chapter 6) . 

Nevertheless, two factors may have contributed to 
somewhat unnatural behavior by caged birds. First, birds 
were subjected to experimental treatments for most or all of 
an entire foraging day. For example, birds were required to 
subsist on limited food or to compete with a cagemate for 
longer than might occur naturally. Consequently, birds may 
have resorted to novel behavior, or they may have experienced 
greater stress than would occur in free-ranging birds. 
Second, during trials with limited food, sucrose solution was 
delivered to feeders in batches at 10-20 min intervals, 
depending on the experiment. Thus, food accumulated in 
feeders as a step function, as opposed to the more natural 
pattern of gradual accumulation described as a ramp function 
(see Lucas and Waser 1989) . Consequently, experimental birds 
typically encountered either an empty feeder (last delivery 
already eaten, next delivery pending) or a relatively large 
meal (delivery just made and not yet eaten) , with 



176 



intermediates unlikely. This pattern resembles the "bonanza- 
blank" nectar distribution described by Feinsinger (1978, 
1983) , which may cause higher visit rates than would an even 
distribution (i.e., a constant but moderate nectar volume per 
flower) . 

Physiological Constraints to Feeding 

Physiological limits to food processing may restrict 
food intake rate and hence influence both behavior and 
energetics of wild birds. Diamond et al. (1986) and Karasov 
et al . (1986) argued that such a physiological constraint on 
feeding rate may explain why hummingbirds spend miuch time 
perching instead of foraging. My first experiment, however, 
indicates that Amazilia and Chlorostilhon typically feed at 
rates that are well below a level constrained 
physiologically. Thus, the hypothesized link between 
digestive constraints and activity (flight vs. perching) is 
weak, and digestive processes are probably not normally 
important determinants of energy budgets for these two 
species . 

Amazilia and Chlorostilhon are not known to be long 
distance migrants. Probably they feed continuously while 
making their short seasonal movements. Thus, these species 
do not encounter the energy demands of many of the temperate- 
tropical m.igrants, such as Selasphorus rufus (Carpenter et 
al . 1983), which make very long distance flights between 



177 



feeding opportunities . Because Selasphorus rufus must store 
much fat during relatively short feeding periods, digestive 
constraints may be important to this species. Nevertheless, 
energy demands for wild Amazilia and Chlorostilhon could be 
unusually high at certain times, such as during reproduction, 
and birds could then be subject to physiological constraints 
on intake. Also, when nectar availability is higher in the 
morning than later in the day, to achieve adequate daytime 
storage these birds may need to process food at extremely 
high rates for part of the day and thus may temporarily face 
digestive constraints to food intake. Nevertheless, 
increased energy demand may itself increase maximal rates of 
food processing (Lee Gass, pers . comm.), so that even when 
demand is high birds may not face physiological constraints 
on intake . Whether any hummingbird species actually achieves 
intake rates in the field that are high enough to be 
constrained by the physiological processes of digestion has 
yet to be determined. 

Intra- and Interspecific Competition 

When Amazilia and Chlorostilhon shared a single 
defensible ad libitum feeder in a simple cage environment, 
birds that appeared behaviorally to be winners were 
energetically more successful than birds that were behavioral 
losers, but both paid appreciable energetic costs. Each 
species exhibited different patterns of energetic costs and 



178 



benefits that may relate to their respective foraging modes . 
Can these conclusions be applied to hummingbirds in their 
natural communities? 

The energetic stress experienced by Chlorostilhon when 
paired with Amazilia suggests a mechanism, operating both 
proximately and ultimately, that may help to explain why a 
single Chlorostilbon is unlikely to share a flower clump with 
a territorial Amazilia if undefended resources are available. 
As a proximate mechanism, Chlorostilbon may depart from 
flower patches defended by Amazilia partly because of low 
rates of energy storage, not simply because it is physically 
excluded from feeding in the patch. Other possible proximate 
mechanisms, such as the inability to obtain food or risk of 
injury to Chlorostilbon, may not adequately explain the 
exclusion of the trapliner from Amazilia territories. Even 
though the single feeder was actively defended by Amazilia, 
Chlorostilbon obtained nearly as much food in heterospecif ic 
pairs as it did when alone. In addition, despite the high 
rates of aggression by Amazilia (up to 4 per min) , no birds 
in any trials sustained visible injury. Behavioral 
mechanisms may also partly determine where Chlorostilbon will 
feed in the wild, with the trapliner tending to avoid patches 
defended by Amazilia because of learned or instinctive 
aversion. Through experience (proximate cause), or 
evolutionary time (ultimate cause) , energetic costs may have 
contributed to the formation of such behavior. 



179 



This experiment, however, was heavily "stacked against" 
the trapliner by simulating extremely clumped resources, 
conditions that are not often encountered in the field. Not 
only was the single feeder quite easy for the dominant 
territorialist to defend against Chlorostilbon, but also the 
smaller trapliner had no other feeding options available 
(e.g., it could not visit undefended low quality "insect" 
flowers) . When Chlorostilbon did break through the feeder 
defense, however, it encountered an unnaturally rich food 
source and could fill its crop completely in just a few 
seconds. The ad libitum feeder may therefore have helped to 
counteract the limitation of feeding opportunities, but the 
cage environment still favored the territorialist . 

This bias for Amazilia is reflected in the 
territorialist ' s consistent energetic advantage over the 
trapliner. Comparing daytime energy storage rates, Amazilia 
performed better than Chlorostilbon under conditions of both 
inter- and intraspecif ic competition. In addition, both 
species suffered less energetic stress when the trapliner 
Chlorostilbon was the cagemate rather than Amazilia . The 
bias favoring Amazilia may persist in the field, but only 
under conditions similar to those in my experiment: an easily 
defensible food source with low intruder pressure. Under 
natural conditions of less defensible resources or greater 
numbers of intruders, the energetic advantage of Amazilia 
could easily diminish. 



180 



The territoriallst may not have competed solely using 
interference. When tested in heterospecif ic pairs Amazilia 
increased both visit rate to feeders and meal size, resulting 
in higher intake than solitary birds. More frequent visits, 
while contributing to higher intake, may also be evidence of 
pre-emptive harvesting (a form of exploitative competition) 
by the territorialist . This combination of exploitative and 
interference competition is analogous to "defense by 
exploitation" (cf. "defense by depletion," Charnov et al . 
1976, Davies and Houston 1981, Waser 1981, Lucas and Waser 
1989) . Paton and Carpenter (1984) documented this phenomenon 
for territorial hummingbirds in the field: birds 
preferentially fed from flowers at the periphery of their 
territories, where intruder pressure was greatest, and thus 
decreased nectar lost to intruders. 

In conspecific pairs, Chlorostilhon increased visit rate 
slightly, as expected for a pre-emptive harvester. 
Phaethornis superciliosus, a high-reward trapliner, when 
competing in the wild with conspecifics (Gill 1988, see 
below) , increased visit rate more markedly than did 
Chlorostilhon in my experiments. Perhaps Chlorostilhon would 
increase visit frequency even more dramatically were more 
limited-food feeders available to visit (see Chapter 7) . 



181 
Food Availability and Interflower Distance 

Birds in trials with LOW food, at either perch distance, 
responded by increasing visits and flight time compared to 
controls. This response, even with DHC, resulted in 
significant loss of energy stores. Birds on LOW food could 
have achieved higher rates of energy storage than they did, 
even without using DHC, simply by decreasing visit rate and 
flight time to the minimum necessary to harvest accumulating 
sucrose solution. If birds had waited 1 h between feeder 
visits, the volume of sucrose solution that accumulated in 
the main feeder prior to each visit would have been less than 
the crop volum.e for each species (510 |IL/510 |IL and 300 
|j,L/436 jlL; feeder volume to crop volume for Amazilia and 
Chlorostilbon, respectively, Hainsworth and Wolf 1972a) . 
This visit rate would require only eight trips to the feeder 
during an 8-h trial, which at 20 s per trip for the FAR 
feeder would result in only about 3 min total flight time 
(i.e., less than 1% flight time) . Such a low flight time 
would result in a mean total energy expenditure very close to 
the mean RMR . Because the mean LOW food delivery rate was 
460 J-g-^-h"^, whereas the mean normal (unconstrained) RMR was 
480 J-g~i-h~i, LOW food birds could have had a m.ean storage 
rate of -20 J-g~^-h~^ by using minimal flight times, as 
opposed to the observed mean rate of -168 J- g~-^ • h"-'- . 

The response of boosting visits and flight time was 
unsuccessful in the cage experiments because additional 



182 



foraging effort produced no reward beyond the limited food 
available in the feeders. In the field, however, even at low 
flower density, additional foraging effort would normally 
allow a bird to increase intake rate, as well as to explore 
and possibly locate richer patches. In addition, both wild 
and captive birds may respond to low food as if it were the 
result of nectar removal by another bird. Reacting as 
exploitative competitors, they may then increase visits and 
harvest pre-emptively — an adaptive response to competitive 
pressure (Gill 1988) . Finally, experimental birds may have 
visited feeders more frequently than the "optimal" 1 visit 
per h at LOW food, because they may have been unable to learn 
the food delivery schedule, or may have continued to revisit 
empty feeders simply because no other food sources were 
available to check. Thus, inappropriate behavior in a 
restrictive cage experiment may represent (a) a successful 
strategy under real conditions, (b) the inability of birds to 
assess the experimental protocol, or (c) the lack of 
alternative foraging opportunities. 

The significant positive correlation between flight time 
and total energy expenditure (see also Chapter 4) indicates 
the importance of flight time to energy management in caged 
hummingbirds, and should apply as well to birds in the field. 
Weathers and Stiles (1989) found no correlation between time 
spent in flight and field metabolic rate (FMR = daily energy 
expenditure) for three nectarivores (two hummingbirds and one 
honeyeater) . Their conclusion was based, however, on a mean 



183 



flight time and mean FMR for each species. Given that flight 
time varies widely among individuals and with different 
foraging conditions, this lack of correlation is not 
surprising. For a given individual, the tim.e spent in flight 
could still be the best predictor of 24-h energy expenditure. 
Weathers and Stiles emphasize "the importance of energy- 
requiring processes other than flight as determinants of FMR" 
(1989, p. 328). Their conclusion is well justified, 
especially for 24-h FMRs in which DHC or overnight torpor 
could easily cancel expenditures due to high flight time. In 
my experiment, variation in the DHC exhibited by each 
treatment group (see Fig. 5-8) may have introduced scatter 
into the regression of flight time vs. energy expenditure for 
each species when treatment groups were combined. The 
regression coefficients increased when data were analyzed by 
food and distance level for each species (mean standardized 
regression coefficients = .900 and .813 for Amazilia and 
Chlorostilbon, respectively) , compared to analysis by species 
alone. In addition, factors such as thermoregulation costs 
might also serve to mask the importance of flight time as a 
determinant of FMR. The above caveats argue for the use of 
caution in estimating energy budgets from time budgets 
(Goldstein 1988) . 

The territorial hummingbird, Amazilia saucerottei, and 
the traplining species, Chlorostilbon canivetii, shared some 
common responses to foraging constraint. When feeders had 
lim.ited food or were far from the perch, birds increased 



184 



foraging expenditures. At the same time, foraging- 
constrained birds apparently reduced resting metabolic rates. 
This enabled them to increase foraging effort, as measured by 
flight time, while maintaining a net energy savings. For 
birds in the wild, DHC might serve to "buy flight time" that 
would allow them to forage at very low food density. 
Additional flight time could also be used to seek out more 
profitable patches. 

Daytime torpor might be a more effective means for 
reducing immediate energetic expenditures, but it presents 
several risks. Torpid birds are more prone to predation 
because they can neither detect nor flee attackers. Torpid 
birds cannot repel intruders from their territories, nor take 
advantage of the departure of a territory holder. Finally, 
torpid birds must expend considerable energy to resume normal 
activity (Hammel et al. 1958). To offset such warm-up costs, 
torpor must be maintained for a fairly long period of time to 
effect a net energy savings . 

In contrast, DHC is less risky than torpor because birds 
remain somewhat alert and responsive. Birds .also have 
reduced warm-up costs for DHC compared to torpor: this 
further reduces risks and allows birds to use DHC 
intermittently between feeding bouts. Although DHC may not 
conserve as much energy as torpor, the birds using DHC in 
this study achieved a relative energy savings of about 25% 
compared to normal RMRs . 



185 



Both species exhibited similar patterns in extraneous 
flight. Although not required for feeder visits, this flight 
time may serve three functions that could explain its 
prevalence when feeders were FAR from the perch. Extraneous 
flight may be used for (a) exploration: to become more 
familiar with the cage layout, (b) sampling: to locate new 
nectar sources for another day, or (c) surveillance: to check 
for territorial intruders or defenders (P. Feinsinger, pers . 
comm.). When feeders are NEAR, all three can be accomplished 
while perched or with short flights. However, when feeders 
are FAR, and hence the flight tube is longer, a bird must 
spend more time in the air to perform these functions. In 
addition, extraneous flight is energetically expensive yet 
may yield no energy in return. Thus, it may be more frequent 
when birds are satiated and hence not needing to feed. In 
support of this idea, extraneous flight was most pronounced 
at HIGH food. 

One interspecific difference appeared counterintuitive. 
When foraging at dispersed low-reward flowers, the trapliner 
Chlorostilbon presumably experiences greater energetic 
success than the territorialist (Feinsinger 1976) . Yet in m^y 
experiment, when foraging at low-reward feeders at either 
distance, Chlorostilbon experienced somewhat reduced 
energetic success compared to Amazllia . Does this indicate 
that Chlorostilbon does not really hold an energetic 
advantage over Amazllia when visiting dispersed low-reward 
flowers in the wild? No, because the little trapliner, with 



186 



its smaller body mass relative to Amaziliar should obtain a 
higher mass-specific intake rate than Amazilia when visiting 
the same numJoer of low-reward flowers. Thus, on a per-bird 
basis, Chlorostilbon can m^eet its intake requirements by 
visiting fewer flowers than Amazilia . In my experiment, 
however, each species was offered the same per-gram level of 
food delivery to the feeders, so the energetic success of 
each species was not tested under identical per-bird food 
availabilities . 

With its strong predisposition toward high energy 
expenditure, Chlorostilbon does not appear prone to "wait 
out" short periods of nectar scarcity. In contrast, the 
territorial Ajnazilia, with its lower flight times and mass- 
specific energy expenditure, appears to be a more successful 
energy manager than Chlorostilbon when food is fixed at a low 
level, as in this experiment. Amazilia, however, may face a 
significant energy challenge when attempting to maintain an 
oversized territory or when traplining very rich but low- 
density nectar sources, such as widely scattered small 
Hamelia bushes . 

Food Dispersion and Interspecific Competition 

Each of the two treatments of feeder dispersion had 
strengths and weaknesses in simulating natural conditions. 
The treatm.ent with clumped feeders was reasonably 
representative of a small feeding territory for Amazilia, 



187 



such as a large bush of Hamelia or an isolated clump of 
Lobelia. The clumped feeders were easily defensible, so that 
during most periods Amazilia restricted feeder access by 
Chlorostilbon to varying extents. Unlike experiment 2 
(single ad libitum feeder. Chapter 4), however, Chlorostilbon 
could feed at one feeder in the clump while Amazilia fed at 
another. This occurs naturally in the field, with intruders 
more likely to enter a territory and obtain a meal when the 
owner is preoccupied with feeding or chasing another intruder 
(Feinsinger 1976, H. Tiebout, pers . obs . and P. Feinsinger, 
pers . comm.) . Also similar to natural conditions, Amazilia 
often defended empty feeders within its cage territory. When 
the trapliner breached the defenses of the territorialist , it 
could find that the feeder (s) it visited was empty, unlike 
experiment 2 . 

No feeders were available to Chlorostilbon, however, 
outside the 5-feeder territory. Thus, the trapliner had no 
choice but to intrude successfully or go hungry. Under 
natural conditions, low densities of flowers would normally 
occur between defensible patches. Thus, the clumiped feeders 
represented the unlikely field situation in which all flowers 
are potentially defensible. Under these conditions, with an 
Amazilia to defend each patch, Chlorostilbon should 
experience lower foraging efficiency and thus lower energy 
storage than Amazilia, forcing the trapliner to emigrate or 
perish over a quite short period of time. This scenario 
assumes at most a single Chlorostilbon intruder per Amazilia 



188 



territory. However,- with additional intruders present, 
trapliners could have more opportunities to feed in 
territories while the owners chased the other intruders. 
Thus, a slight increase in intruder pressure might favor 
Chlorostllbon somewhat, though such benefits would soon fall 
as increasing intruder pressure caused the total standing 
crop of nectar to be divided among more birds . Under even 
more complex natural conditions, involving multiple species 
with a diversity of foraging modes, the relative energetic 
successes of Amazilia and Chlorostllbon sharing clumped 
resources would be increasingly difficult to predict. 

The treatment with dispersed feeders only roughly 
approximated natural low density flower conditions. While 
the dispersed feeders were not defensible, they provided more 
energy than birds might typically obtain from a single flower 
in the field. Successive feeders were only 1 to 8 m apart, 
and only 4 m from the nearest perch. Thus, flight time did 
not increase significantly (dispersed > clumped by 18%; 
P > 0.1) and averaged only 25.7%, compared to 53.4% when 
perch-to-feeder distance was 20 m (see Chapter 5) . The 
relatively small amount of time spent in flight may help 
explain why the lower wing disc loading of Chlorostllbon did 
not translate into appreciable energy savings for the 
trapliner (but see discussion in Chapter 5) . If birds faced 
much greater interfeeder distances than were used in this 
current experim.ent, perhaps the differences in wing disc 



189 



loading would give Chlorostilhon an energetic advantage over 
Ama zllia . 

The low storage rates of Chlorostilhon, compared to 
Amazilia, when feeders were dispersed may be due in part to 
the limited number of feeders available. Chlorostilhon had a 
somewhat higher intake rate per feeder visit than Amazilia, 
and thus in one sense was the better harvester. Because of 
the trapliner's greater intake rate requirement, however, it 
apparently cannot handle absolute food limitation (as opposed 
to limitation due to low food density) as well as Amazilia . 
Thus, to meet its high energy requirements, the trapliner 
must couple its harvesting advantage with the opportunity to 
visit more feeders. Based on its performance in a cage, 
Chlorostilhon appears to be a high intake rate, low 
profitability forager compared to Amazilia. When only low- 
reward flowers are available in the wild, however, 
Chlorostilhon may have an energetic edge over Amazilia 
because of the smaller body mass of the trapliner, as 
previously discussed. 

Regardless of feeder dispersion, both species tended to 
decrease visit diversity and feeder overlap over the foraging 
day. These changes in resource utilization patterns 
presumably reduced the energetic consequences of competition 
and thus enhanced foraging profitability. Under natural 
conditions, with many flower species available at varying 
patterns of dispersion, would these trends persist? 



190 



Certainly hummingbirds should adjust their patterns of 
resource utilization to reduce costs and enhance storage 
rates, if possible, but this may not always involve 
decreasing both visit diversity and visit overlap. If food 
is abundant to moderately limited, then birds will profit by 
reducing overlap with potential competitors (analogous to low 
visit diversity) . However, if food is relatively scarce, 
then birds may be forced to overlap greatly and to visit 
every possible flower (analogous to high visit diversity) . 
Furthermore, overlap and diversity in use of resources need 
not vary together, nor need species with different foraging 
modes respond similarly to given resource conditions. For 
example, when Amazllia successfully excludes Chlorostilbon 
from rich Hamelia bushes, Chlorostilbon may trapline a large 
number of low-reward flowers of many species. Thus, Amazllia 
would have low visit diversity while Chlorostilbon had high 
diversity, with low flower overlap between them. With mixed 
flower dispersion, then, the foraging pattern shown in the 
cage experiments — i.e., both species reducing visit 
diversi"cy — would not likely hold. Patterns of overlap can 
become increasingly complex when both intra- and 
interspecific interactions are considered, and the addition 
of more species will cause a geometric increase in 
complexity . 

The DLW experiment demonstrated that individual birds 
can make short-term changes in visit diversity and feeder 
overlap when sharing food, and that these responses enhance 



191 



profitability. A similar phenomenon has been reported for 
traplining hermit hummingbirds (Phaethornis superciliosus) 
competing intraspecif ically (Gill 1988) . Some birds timed 
visits to harvest nectar before competitors (see Chapter 5) 
and thus gained control of individual flowers as competitors 
visited them less and less frequently. Ultimately, birds 
tended to segregate among undefended feeding sites, with 
exclusive use apparently resulting from previously successful 
pre-emptive harvesting. This process of dividing resources, 
and ultimately reducing costs, represents another means of 
energy managem^ent available to both territorialists and 
trapliners . 

General Trends Across Experiments 

The effects on foraging birds of the five determinants 
of energetic success (foraging mode, food processing 
constraints, competition, food availability, and food 
dispersion) were often dependent upon context. For example, 
the advantage of a given foraging m.ode depended, upon food 
availability and the foraging mode of the cagemate . In 
addition, there were many significant statistical 
interactions between or among factors within a single 
experiment. For example, the time between meals showed a 
significant three-way interaction among species, competition 
type, and number of birds sharing the feeder (Chapter 4) . 
Such complexity makes it difficult, and of questionable 



192 



utility, to distill generalities about the effects of any 
individual factor. Several patterns resulting from 
combinations of factors, however, occurred throughout my 
experiments or across treatment groups and thus appear 
robust. These patterns, discussed below, may be likely to 
occur as well under natural conditions . 

Wild birds may experience several energetic consequences 
to sharing a food source. When food is shared, regardless of 
competition type, it must be divided among participants; yet 
at the same time, sharing food tends to increase the 
energetic demands of participants by increasing expenditures. 
This increased demand for food interacts with food 
availability in situation-specific ways to determine whether 
a bird can meet its daytime energy storage requirements. For 
example, when the shared food source is defensible, a 
resident Amazilia that limits access by intruders may be able 
to increase intake rates sufficiently to meet increased 
expenditures, thus ensuring adequate daytime storage. This 
energetic success can occur when defending against hetero- or 
conspecifics, and at abundant or limited food. In contrast, 
both the trapliner Chlorostilbon and subordinate Amazilia can 
achieve energetic success only when paired with Chlorostilbon 
and only at abundant food. 

Several interspecific differences remain relatively 
invariant and persist whether birds were tested in pairs or 
alone. The most conspicuous is that Chlorostilbon typically 
exhibits both higher RMRs and higher rates of overall energy 



193 



expenditure than does Amazilia . As previously discussed, 
this results from higher BMRs for the trapliner, as predicted 
by its smaller body mass, as well as greater levels of 
activity (primarily flight) . The effects of BMR and activity 
are linked via positive feedback, because a high BMR requires 
high food intake. By specializing on low-reward and 
dispersed flowers, Chlorostilhon may typically maintain low 
foraging profitability, compared to a territorial species. 
Its high overhead (BMR) requires high foraging expenses, 
which further increase food demand. Thus, the little 
trapliner must maintain high mass-specific intake rates 
relative to the territorialist . Because it has a very low 
energy storage capacity (Paladino 1989), Chlorostilhon, like 
other small endotherms (McNab 1989, p. 287), may need to 
spend most of its day foraging. 

Given the large amount of time a low-reward trapliner 
may have to spend in foraging flight, selection should favor 
a mechanism to reduce these costs. This mechanism is 
apparent in the relatively low wing disc loading of 
Chlorostilbon . Nevertheless, results from both a small cage 
(all hovering) and a large cage (mixed forward flight and 
hovering) failed to detect a significant energetic difference 
between species that is attributable to differences in flight 
cost per unit time. Possible explanations for this are 
presented in Chapter 5. However, even if Chlorostilbon did 
exhibit somewhat lower flight costs per unit time than 
Amazilia, the interspecific differences in these costs still 



194 



could be rendered relatively unimportant in terms of overall 
energy budgets, because Chlorostilbon tends to spend so much 
more time flying than Amazllla. Additional studies are 
needed with larger sample sizes and closer monitoring of 
short-term changes in actual body mass to detect the 
energetic consequences of wing disc loading, which may be 
important over ecological and evolutionary time (Feinsinger 
and Chaplin 1975, Feinsinger and Colwell 1978, Feinsinger et 
al. 1979) . 

Finally, birds tend to diminish activity and 
expenditures over the course of a foraging day, despite the 
constant availability of food (whether ad libitum or 
limited) . When food is shared, this reduction may reflect 
partitioning of limited food resources, which allows greater 
profitability (Chapter 7), but reduction in activity and 
expenditures also occurs when food is neither limited nor 
shared (Chapter 5) . Wild Amazllla and Chlorostilbon may 
exhibit a similar temporal response as an adaptation to 
natural patterns of nectar availability at the Monteverde 
site (P. Feinsinger, pers. comm. ) . Many of the commonly 
visited flower species secrete the most nectar in the 
morning, tapering off in the afternoon (Feinsinger 1976, 
1978) . Reduced activity in the afternoon has also been noted 
for hummingbirds at other neotropical sites (Stiles 1975) and 
may in fact be typical for most bird species (Kendeigh et al. 
1977) . Whether this pattern in hummingbirds reflects reduced 
food availability, reduced competition due to partitioning of 



195 



resources (Gill 1988)', or reduced food demand (Kendeigh et 
al . 1977) is still open to question. 

The Mechanistic Approach to Understanding Natural Guilds 

In this study I investigated the roles of several 
factors believed to be energetically important to foraging 
hummingbirds. Each factor was examined in one or more simple 
experiments under very restricted conditions. To be able to 
extrapolate with any confidence from these types of 
experiments to natural conditions would first require a 
series of additional experiments. Amazilia and Chlorostilbon 
need to be tested under conditions involving a more complete 
range of feeding opportunities designed to sim.ulate the full 
extent of natural conditions. For example, experiments in 
very large flight cages could provide birds with both clumped 
and dispersed feeders, or a mixture of high and low reward 
feeders, thereby allowing foragers to choose among available 
resource states . Outdoor aviaries could be used to examine 
the influence of seasonal climatic changes that may influence 
foraging profitability. For example, the high winds and 
accompanying mist that Monteverde experiences from October to 
January each year may be more stressful energetically to 
small weak-flying hum.mingbirds such as Chlorostilbon than to 
larger and faster species such as Amazilia . 

The results of my experiments indicate that the most 
important factor yet to be investigated is the 



196 



intake : expenditure ratio, when both energy terms are free to 
vary. Experiments in which both intake and expenditure 
varied with foraging effort have recently been conducted on 
the domestic pigeon, Columba livia. Food costs were 
indirectly controlled by requiring a specified number of 
pecks before presentation of a food hopper (Rashotte and 
Henderson 1988, Ostheim and Rautenberg 1989, Rashotte et al. 
1989) . Although these particular experiments were performed 
on restrained birds and hence may not extrapolate well to 
natural foraging, this paradigm of the "closed economy" 
(Hursh 1980, 1984; Collier and Rovee-Collier 1981, Collier 
1983) should provide an important conceptual framework for 
more natural experiments (Houston and McNamara 198 9) . For 
example, hummingbirds could be tested with food limited by 
its density (i.e., how far a bird m.ust fly per unit intake) 
rather than by the absolute amount of food available (i.e., a 
fixed low level of food that is unaffected by foraging 
effort) . When food is limited by the amount of energy 
required to obtain it, rather than absolutely (as in my 
experiments) , birds with low foraging efficiency (such as 
Chlorostilbon) should achieve greater daily storage by 
foraging longer. Such experiments may ultimately reveal 
important differences between species based on their foraging 
efficiencies per se, not simply on their ability to tolerate 
low food levels . 

Even with a wide variety of paired-bird experiments 
there will be difficulties in generalizing to the natural 



197 



communities these birds inhabit. In the Monteverde 
successional habitats where Amazilia and Chlorostilbon are 
common, they may regularly interact with 10 additional 
trochilid species and irregularly with yet another six 
(Feinsinger 1976, H. Tiebout unpublished data) . The other 
species utilize a variety of foraging modes in addition to 
territoriality and low-reward traplining. These other 
species also span a wider range of body size and bill 
morphology than do Amazilia and Chlorostilbon, with all of 
the implications for resource utilization and competition 
these characteristics entail. In addition to interacting 
with other Trochilidae, hummingbirds also frequently compete 
for nectar with flower-piercers (Colwell et al . 1974) and 
various insects (Boyden 1978, Carpenter 1979, Gill et al. 
1982, Laverty and Plowright 1985) . 

Furthermore, hummingbird species are not rigidly cast in 
their foraging roles (Feinsinger and Colwell 1978) . In my 
experiments Amazilia was capable of foraging profitably at 
dispersed, limited resources, and Chlorostilbon exhibited 
interference competition under certain conditions. Foraging 
behavior can vary altitudinally within a single hummingbird 
species, possibly in response to energetic factors (Berger 
1974a, 1974b; Feinsinger et al . 1979). Similarly, at a 
single site the foraging mode of an individual bird may be 
somewhat context-specific, depending upon food availability 
and dispersion, and qualitative and quantitative 






198 



characteristics of competitor pressure (Feinsinger 1976, 
Feinsinger and Colwell 1978) . 

In addition, humraingbirds do not eat only nectar, and 
the contributions to their energy budgets from arthropods 
(Montgomerie and Redsell 1980, Remsen et al. 1986, Parrish 
1987), fruit (Smith 1926), and sap (Southwick and Southwick 
1980) have been little studied. The importance of 
competition for these food sources is unknown. Even within 
the nectar food base, sugar composition varies among flower 
species, yet we currently know little about species-specific 
capabilities of hummingbirds for digesting various mono- and 
disaccharides . The energetic importance of other nectar 
constituents, such as proteins and lipids (Baker and Baker 
1975), remains virtually unstudied. 

Finally, as mentioned in Chapter 1, energetic success is 
but one of many possible determinants of a species' presence 
or absence in a given area. Energetic success is necessary 
but not sufficient to explain a species' presence, and thus 
its predictive value is dependent upon understanding the 
other determinants as well. In addition, the relative 
importance of a given determinant may be dependent upon 
events occurring elsewhere. Birds are highly vagile and can 
respond rapidly to spatial variation in food resources by 
emigration and immigration. A species' representation in a 
guild need not depend only on the characteristics of a 
particular site, but may instead reflect the quality of that 
site relative to other available sites. 



199 



Despite the many difficulties in extrapolating from 
simplistic experimients to natural communities, such efforts 
may ultimately yield greater predictive power (Tilman 1987) 
than the currently popular phenomenological approach to 
species interactions. Furthermore, such experiments can 
reveal the species-specific mechanisms responsible for 
creating the patterns we observe at the level of guild and 
above (Schoener 1986, Tilman 1987) . The four major 
experiments that comprise my study are a first step towards 
assessing the importance of one such mechanism, relative 
energetic success, as a determinant of hummingbird guild 
composition. I have provided evidence that the hummingbird 
species I studied are not typically limited in their feeding 
by digestive constraints. Thus, limits to intake in the wild 
must normally be imposed ecologically, through either limited 
food availability, competition for food, inclement weather, 
or a combination of such factors . The importance of 
ecological limits suggests that differences in the ability of 
hummingbirds to acquire food, as determined in part by 'their 
variation in foraging modes, can have energetic consequences. 

Differential energetic success was demonstrated in the 
last three experiments I conducted with Amazllia and 
Chlorostilbon . These species differed, in both their 
relative foraging profitabilities and in the flexibility of 
their energy budgets, in response to the ecological factors I 
manipulated. Thus, species representing two foraging modes 
differed energetically in ways indicating that each was 



200 



specialized for particular patterns of food availability and 
flower dispersion. 

The central proposition of my study was that guild 
structure could be determined by differential energetic 
success among species of various foraging modes. My goal was 
to determine the validity of one necessary precondition of 
this proposition: that species with different foraging modes 
do in fact differ energetically under varying foraging 
conditions. The results of my study confirm this 
precondition, and set the stage to extend research on the 
role of energetics as a determinant of guild dynamics. 

Future research must first demonstrate a link between 
interspecific energetic differences and such differences 
among foraging modes as groups . Because my study tested only 
a single representative territorialist and low-reward 
trapliner, I cannot yet make such inferences with confidence. 
Second, a link between energetic differences among species 
and their consequent representation in a local guild must be 
documented under natural conditions with free-ranging birds. 
By investigating this energetic mechanism in a diversity of 
other guildmates, as well as other mechanisms important to 
these animals, we may gradually build an understanding of the 
processes responsible for the highly dynamic structure of 
tropical nectarivore assem.blages . 



LITERATURE CITED 



Austin, G.T. 1970. Interspecific territoriality of migrant 
calliope and resident broad-tailed hummingbirds. Condor 
72: 234. 

Baker, H.G., and I. Baker. 1975. Studies of nectar- 
constitution and pollinator-plant coevolution. pp. 100-140 
In: L.E. Gilbert and P.H. Raven (Eds.) Coevolution of 
animals and plants. University of Texas Press, Austin, TX . 

Bartholomew, G.A., and T.M. Casey. 1978. Oxygen consumption 
of moths during rest, preflight warmup and flight in 
relation to body size and wing morphology. Journal of 
Experimental Biology 76:11-25. 

, T.A. Howell, and I.J. Cade. 1957. Torpidity in the 



white-throated swift, Anna hummingbird and poor-will. 
Condor 59: 145-155. 

Bennett, A.F., and L.D. Houck . 1983. The energetic cost of 
courrship and aggression in a plethodontid salamander. 
Ecology 64:979-983. 

Berger, M. 1974a. Energiewechsel von Kolibris beim 
Schwirrf lug unter Hohenbedingungen . Journal fur 
Ornithologie 115:273-288. 

. 1974b. Oxygen consumption and power of hovering ' 

hummingbirds at varying barometric and oxygen pressures 
Naturwissenschaften 61:407. 

. 1978. Ventilation in the hummingbird, Colihri 



coruscans , during altitude hovering, pp. 85-88, In: J. 
Piper (Ed.) Respiratory function in birds: adult and 
embryonic. Springer-Verlag, New York. 

, and J.S. Hart. 1972. Die Atmung beim Kolibri Amazilia 

fimbrlata wahrend des Schwirrf luges bei verschiedenen 
Umgebungstemperaturen . Journal of Comparative Physiology 
81:363-380. 

, and . 1974. Physiology and energetics of flight. 

Chapter 5 In: D.S. Earner and J.R. King (Eds.) Avian 
biology, vol. IV. Academic Press, New York, 504 pp. 



201 



202 



Beuchat, C.A., S.B. Chaplin, and M.L. Morton. 1979. Ambient 
temperature and the daily energetics of two species of 
hummingbirds, Calypte anna and Selasphorus rufus. 
Physiological Zoology 52: 280 - 295. 

Birch, L.C. 1957. The meanings of competition. American 
Naturalist 91:5-18. 

Birt-Friesen, V.L., W.A. Montevecchi, D.K. Cairns, and S.A. 
Macko . 1989. Activity-specific metabolic rates of free- 
living northern gannets and other seabirds . Ecology 70: 
357-367. 

Bligh, J., and K.G. Johnson. 1973. Glossary of terms for 

thermal physiology. Journal of Applied Physiology 35:941- 
961. 

Bolten, A.B., P. Feinsinger, H.G. Baker, and I. Baker. 1979. 
On the calculation of sugar concentration in flower 
nectar. Oecologia 41:301-304. 

Boyden, T.C. 1978. Territorial defense against hummingbirds 
and insects by tropical hummingbirds. Condor 80:216-221. 

Brown, J.H., and M.A. Bowers. 1985. Community organization in 
hummingbirds: relationships between morphology and 
ecology. Auk 102: 251 - 269. 

, W.A. Calder III, and A. Kodric-Brown . 1978. Correlates 



and consequences of body size in nectar-feeding birds. 
American Zoologist 18: 687-700. 

, A. Kodric-Brown, T.G. Whitham, and H.W. Bond. 1981. 



Competition between hummingbirds and insects for the 
nectar of two species of shrubs. The Southwestern 
Naturalist 26:133-145. 

Bucher, T.L., and M.A. Chappell. 1989. Energy metabolism and 
patterns of ventilation in euthermic and torpid 
hummingbirds. In: C. Bech and R.E. Reinertsen (Eds.) 
Physiology of cold adaptation in birds. NATO ASI Series, 
Vol. 173, Plenum Press, New York. 

Calder, W.A., III. 1974. Consequences of body size for avian 
energetics. In: R.A. Paynter (Ed.) Avian energetics. 
Publication of the Nuttall Ornithology Club No. 15, pp. 
86-151 . 

Carpenter, F.L. 1974. Torpor in an Andean hummingbird -- its 
ecological significance. Science 183: 545-547. 

. 1976. Ecology and evolution of an Andean hummingbird. 

University of California Publications in Zoology, Vol. 
106, University of California, Berkeley, CA. 



203 



. 1979. Competition between hummingbirds and insects for 

nectar. American Zoologist 19:1105-1114. 

, M.A. Hixon, and D.C. Paton. 1981. Premigratory changes 

in time budgets of territorial hummingbirds. American 
Zoologist 21:921. 

, and R.E. MacMillen. 1976a. Energetic cost of feeding 

territories in an Hawaiian honeycreeper . Oecologia 26: 
213-223. 

, . 1976b. Threshold model of feeding 



territoriality and test with a Hawaiian honeycreeper. 
Science 194: 639-641. 

, D.C. Paton, and M.A. Hixon. 1983. Weight gain and 



adjustment of feeding territory size in migrant 
hummingbirds. Proceedings of the National Academy of 
Science 80:7259-7263. 

Case, T.J., and M.E. Gilpin. 1974. Interference competition 
and niche theory. Proceedings of the National Academy of 
Science 71:3073-3077. 

Charnov, E.L., G.H. Orians, and K. Hyatt. 1976. Ecological 
implications of resource depression. American Naturalist 
110: 247-259. 

Chew, V. 1976. Comparing treatment means: a compendium. 1976. 
HortScience 11: 348-357. 

Cody, M.L. 1968. Interspecific territoriality am.ong 
hummingbird species. Condor 70:270-271. 

Collier, G.H. 1983. Life in a closed economy: the ecology of 
learning and motivation. In: M.D. Zeiler and P. Harzen 
(Eds.) Advance in analysis of behavior. Vol. 3. Wiley, New 
York. 

Collier, G.H., and C.K. Rovee-Collier . 1981. A comparative 
analysis of optimal foraging behavior: laboratory 
simulations . In: A.C. Kamil and T.D. Sargent (Eds.) 
Foraging behavior: ecological, ethological and 
psychological approaches. Garland Press, New York. 

Collins, E.G., and D.C. Paton. 1989. Consequences of 

differences in body mass, wing length and leg morphology 
for nectar feeding birds. Australian Journal of Ecology. 
14: 269-289. 

Colwell, R.K. 1973. Competition and coexistence in a simple 
tropical community. American Naturalist 107: 737-760. 



204 



B.J. Betts, P. Bunnell, F.L. Carpenter, and P. 



Feinsinger. 1974. Competition for the nectar of 
Centropogon valerii by the hummingbird Colihri thalassinus 
and the flower-piercer Dlglossa plumbea and its 
evolutionary implications. Condor 76: 447-452. 

Copenhaver, C., and P. Ewald. 1980. Cost of territory 

establishment in hummingbirds. Oecologia 45: 155-160. 

Davies, N.B., and A.I. Houston. 1981. Owners and satellites: 
the economics of territory defense in the Pied Wagtail, 
Motacllla alba. Journal of Animal Ecology 50: 157-180. 

Day, R.W., and G.P. Quinn . 1989. Comparisons of treatments 
after an analysis of variance in ecology. Ecological 
Monographs 59: 433-463. 

DeBenedictis, P. A., F.B. Gill, F.R. Hainsworth, G.H. Pyke, 
and L.L. Wolf. 1978. Optimal meal size in hummingbirds. 
American Naturalist 112: 301-316. 

Des Granges, J.L. 1979. Organization of a tropical nectar 

feeding bird guild in a variable environment . The Living 
Bird. 17th Annual: 199-236. 

Diamond, J.M., W.H. Karasov, D. Phan, and F.L. Carpenter. 
1986. Digestive physiology is a determinant of foraging 
bout frequency in hummingbirds. Nature 320: 62-63. 

Dixon, W.J. (Ed.). 1985. BMDP statistical software. 

University of California Press, Berkeley, x + 734 pp. 

Eckert, R., and D. Randall. 1983. Animal physiology, 

mechanisms and adaptations. W.H. Freeman and Co., New 
York, xviii+830 pp. 

Epting, R.J. 1975. Power input for hovering flight in 

hummingbirds and its effect on the time and energy budgets 
of foraging. Ph.D. Thesis, University of California, ' 
Berkeley . 

Epting, R.J. 1980. Functional dependence of the power for 
hovering on wing disc loading in hummingbirds . 
Physiological Zoology 53:347-357. 

Epting, R.J., and T.M. Casey. 1973. Power output and wing 

disc loading in hovering hummingbirds. American Naturalist 
107: 761-765. 

Ewald, P.W., and F.L. Carpenter. 1978. Territorial responses 
to energy manipulations in the Anna hummingbird. Oecologia 
31: 277-292. 



205 



, and G.H. Orians . 1983. Effects of resource depression 

on use of inexpensive and escalated aggressive behaviors: 
experimental tests using Anna hummingbirds. Behavioral 
Ecology and Sociobiology 12: 95-101. 

, and W.A. Williams. 1982. Function of the bill and 

tongue in nectar uptake by hummingbirds. Auk 99: 573-576. 

Feder, M.E., and S.J. Arnold. 1982. Anaerobic metabolism and 
behavior during predatory encounters between snakes 
{Thamnophis elegans) and salamanders (Plethodon jordani) . 
Oecologia 53: 93-97. 

Feinsinger, P. 1974. Organization of a tropical guild of 
nectarivorous birds. Ph. D. Thesis, Cornell University, 
Ithaca, NY. 

. 1975. Organization of a tropical guild of 



nectarivorous birds. Ecological Monographs 46: 257-291. 

. 1977. Notes on the hummingbirds of Monteverde, 

Cordillera de Tilaran, Costa Rica. Wilson Bulletin 89: 
159-164. 

. 1978. Ecological interactions between plants and 

hummingbirds in a successional tropical community. 
Ecological Monographs 48: 269-287. 

. 1980. Asynchronous migration patterns and the 

coexistence of tropical hummingbirds. In: Keast A., and A. 
S. Morton (Eds.), Migrant birds in the neotropics: 
ecology, behavior, distribution and conservation, 
Smithsonian Institution Press, Washington, D.C. 

. 1983. Variable nectar secretion in a Heliconia species 



pollinated by hermit hummingbirds. Biotropica 15: 48-52. 

, J.H. Beach, Y.B. Linhart, W.H. Busby, and K.G. Murray. 

1987. Disturbance, pollinator predictability, and 
pollinator success among Costa Rican cloud forest plants. 
Ecology 68: 1294-1305. 

, W.H. Busby, K.G. Murray, J.H. Beach, W.Z. Pounds, and 

Y.B. Linhart. 1988. Mixed support for spatial 
heterogeneity in species interactions: hummingbirds in a 
tropical disturbance mosaic. 

, and S.B. Chaplin. 1975. On the relationship between 

wing disc loading and foraging strategy in hummingbirds. 
American Naturalist 109: 217-224. 

, and R.K. Colwell. 1978. Community organization among 

neotropical nectar-feeding birds. American Zoologist 18: 
779-793. 



206 



, , J. Terborgh, and S.3. Chaplin. 1979. Elevation 

.and the morphology, flight energetics and foraging ecology 
of tropical hummingbirds. American Naturalist 113: 481- 
497 . 

, L.A. Swarm, and J. A. Wolfe. 1985. Nectar-feeding 

birds on Trinidad and Tobago: comparison of diverse and 
depauperate guilds. Ecological Monographs 55: 1-28. 

Gass, C.L., and K.P. Lertzman. 1980. Capricious mountain 
weather: a driving variable in hummingbird territorial 
dynamics. Canadian Journal of Zoology 58: 1964-1968. 

, and R.D. Montgomerie. 1981. Hummingbird foraging 



behavior: decision-making and energy regulation. In: A.C. 
Kamil and T.D. Sargent (Eds.) Foraging behavior: 
ecological, ethological and psychological approaches. 
Garland Press, New York. 

Gill, F.B. 1978. Proximate costs of competition for nectar. 
American Zoologist 18: 753-763. 

. 1985. Hummingbird flight speeds. Auk 102: 97-101. 



1988. Trapline foraging by hermit hummingbirds: 



competition for an undefended, renewable resource. 
Ecology 69: 1933-1942. 

, A.L. Mack, and R.T. Ray. 1982. Competition between 



hermit hummingbirds (Phaethorninae) and insects for nectar 
in a Costa Rican rain forest. Ibis 124: 44-49. 

Goldstein, D.L. 1988. Estimates of daily energy expenditure 
in birds: the time-energy budget as an integrator of 
laboratory and field studies. American Zoologist 28: 829- 
844. 

Greenewalt, C.C. 1975. The flight of birds. Transactions of 
the American Philosophical Society 65:1-67. 

Hails, C.J. 1979. A comparison of flight energetics in 

hirudines and other birds. Comparative Biochemistry and 
Physiology 63A: 581-585. 

Hainsworth, F.R. 1973. On the tongue of a hummingbird: its 
role in the rate and energetics of feeding. Comparative 
Biochemistry and Physiology 46A: 65-78. 

. 1974. Food quality and foraging efficiency: the 

efficiency of sugar assimilation by hummingbirds. Journal 
of Comparative Physiology 88: 425-431. 

^ . 1978. Feeding: models of costs and benefits in energy 



regulation. American Zoologist 18: 701-714 



207 



, 1981a. Animal physiology: adaptations in function 



Addison-Wesley Publishing Co. Reading, MA. xiii + 669 pp. 

. 1981b. Energy regulation in hummingbirds. American 

Scientist' 69: 420-428. 

, B. G. Collins, and L. L. Wolf. 1977. The function of 

torpor in hummingbirds. Physiological Zoologist 50: 215- 
222. 

, T. Mercier, and L.L. Wolf. 1983. Floral arrangements 

and hummingbird feeding. Oecologia 58: 225-229. 

, M.F. Tardiff, and L.L. Wolf. 1981. Proportional 

control for daily energy regulation in hummingbirds . 
Physiological Zoology 54: 452-462. 

, and L.L. Wolf. 1970. Regulation of oxygen consumption 



and body temperature during torpor in a hummingbird, 
Eulampls jugularis . Science 168: 368-359. 

, and . 1972a. Crop volume, nectar concentration 

and hummingbird energetics. Comparative Biochemistry and 
Physiology 42A: 359-366. 

, and . 1972b. Energetics of nectar extraction in a 

small, high altitude tropical hummingbirds, Selasphorus 
flammula . Journal of Comparative Physiology 80: 377-387. 

, and . 1972c. Power for hovering flight in 

relation to body size in hummingbirds. American Naturalist 
106: 589-596. 

, and . 1975. Wing disc loading: implications and 



importance for hummingbird energetics. American Naturalist 
109: 229-233. 

, and . 1976. Nectar characteristics and food 

selection by hummingbirds. Oecologia 25: 101-113. 

, and . 1983. Models and evidence for feeding 



control of energy. American Zoologist 23: 261-272. 

Hammel, H.T., T.J. Dawson, R.M. Abrams,' and H.T. Andersen. 
1968. Total calorimetric measurements on Citellus 
lateralis in hibernation. Physiological Zoology 41: 341- 
357. 

Heinrich, B. 1970. Thoracic temperature stabilization in a 
free-flying moth. Science 168: 580-582. 



208 



1971. Temperature regulation of the sphinx moth/ 



Manduea sexta . I. Flight energetics and body temiperature 
during free and tethered flight . Journal of Experimental 
Biology 54: 141-152. 

. 1974. Thermoregulation in endothermic insects. Science 

185: 747-755. 

. 1975. Thermoregulation in bumblebees. II. Energetics 

of warm-up and free flight. Journal of Comparative 
Physiology 96: 155-166. 

. 1977. Why have some animals evolved to regulate a high 

body temperature? American Naturalist 111: 623-640. 

, and G.A. Bartholomew. 1971. An analysis of pre-f light 



warm-up in the sphinx moth, Manduea sexta. Journal of 
Experimental Biology 55: 223-239. 

Heller, H.C. 1989. Sleep, hypometabolism, and torpor in 

birds. In: C. Bech and R.E. Reinertsen (Eds.) Physiology 
of cold adaptation in birds. NATO ASI Series, Vol. 173, 
Plenum Press, New York. 

Hixon, M A., F.L. Carpenter, and D.C. Paton. 1983. Territory 
area, flower density, and time budgeting in hummingbirds: 
an experimental and theoretical analysis. American 
Naturalist 122: 366-391. 

Hogstad, 0. 1987. It is expensive to be dominant. Auk: 333- 
336. 

Houston, A. I., and J.M. McNamara . 1989. The value of food: 
effects of open and closed economies. Animal Behavior 37: 
546-562. 

Hursh, S.R. 1980. Economic concepts for the analysis of 
behavior. Journal of Experimental Analysis Behavior 37: 
219. 

. 1984. Behavioral economics. Journal of Experimental 



Analysis Behavior 42: 435. 

Jaeger, R.G., K.C.B. Nishikawa, and D.E. Barnard. 1983. 
Foraging tactics of a terrestrial salamander: costs of 
territorial defense. Animal Behavior 31: 191-198. 

Johnsgard, P. A. 1983. The hummingbirds of North America. 
Smithsonian Institution Press, Washington, D.C. 303 pp. 

Karasov, W.H., D. Phan, J.M. Diamond, and F.L. Carpenter. 
1986. Food passage and intestinal nutrient absorption in 
hummingbirds. Auk 103: 453-464. 



209 



Kendeigh, S.C., V.R. Dol'nik, and V.M. Gavrilov. 1977. -Avian 
energetics. In: Pinowski, J. and S.C. Kendeigh (Eds.) 
Granivorous birds in ecosystems, Cambridge University 
Press, Cambridge, 127-204. 

Kingsolver, J.G., and T.L. Daniel. 1983. Mechanical 

determinants of nectar feeding strategy in hummingbirds: 
energetics, tongue morphology and licking behavior. 
Oecologia 60: 214-226. 

Kirkwood, J.K. 1983. A limit to metabolizable energy intake 
in mammals and birds . Comparative Biochemistry and 
Physiology 75A: 1-3. 

Kodric-Brown, A., J. H. Brown, G. S. Byers, and D. F. Gori . 
1984. Organization of a tropical island community of 
hummingbirds and flowers. Ecology 65: 1358-1368. 

Kriiger, K., R. Prinzinger, and K.-L. Schuchmann. 1982. 
Torpor and metabolism in hummingbirds. Comparative 
Biochemistry and Physiology 73A: 679-689. 

Lasiewski, R.C. 1963. Oxygen consumption of torpid, resting, 
and flying hummingbirds. Physiological Zoology 36: 122- 
140. 

, and W.R. Dawson. 1967. A re-examination of the 

relation between standard metabolic rate and body weight 
in birds. Condor 69: 13-23. 

Laverty, T.M., and R.C. Plowright. 1985. Competition between 
hummingbirds and bumblebees for nectar in flowers of 
Impatiens bi flora. Oecologia 66: 25-32. 

LeFebvre, E.A. 1964. The use of D2-'-^0 for measuring energy 
metabolism in Columba livia at rest and in flight. Auk 81; 
403-416. 

Lifson, N., and R.M. McClintock. 1966. Theory of use of the 
turnover rates of body water for measuring energy and 
material balance. Journal of Theoretical Biology 12: 46- 
74. 

Lucas, J.R., and P.M. Waser. 1989. Defense through 

exploitation: a Skinner box for tropical rainforests. 
Trends in Ecology and Evolution 4: 62-63. 

Lyon, D.L. 197 6. A. montane hummingbird territorial system in 
Oaxaca, Mexico. Wilson Bulletin 88: 280-299. 

MacArthur, R.H. 1972. Geographical ecology: patterns in the 
distribution of species. Harper and Row, New York, 269 pp 



210 



MacMillen, R.E., and F.L. Carpenter. 1977. Daily energy- costs 
and body weight in nectarivorous birds. Comparative 
Biochemistry and Physiology 56A: 439-441. 

Malloy, L.G., and C.F. Herreid II. 1979. Energetics of 

behavior in single and paired mice. American Zoologist 19: 
991. 

McNab, B.K. 1988. Food habits and the basal rate of 
metabolism in birds. Oecologia 77: 343-349. 

. 1989. Body mass, food habits, and the use of torpor in 



birds. In: C. Bech and R.E. Reinertsen (Eds.) Physiology 
of cold adaptation in birds. NATO ASI Series, Vol. 173, 
Plenum Press, New York. 

Merker, G. P., and K. A. Nagy . 1984. Energy utilization by 
free-ranging Sceloporus virgatus lizards. Ecology 65: 575- 
581. 

Miller, R.S. 1967. Pattern and process in competition. 
Advances in Ecological Research 4: 1-74. 

, and C.L. Gass. 1985. Survivorship in hummingbirds: is 



predation important? Auk 102: 175-178. 

Montgomerie, R.D. 1979. The energetics of foraging and 

competition in some Mexican hummingbirds. Ph. D. Thesis, 
McGill University, Montreal. 

. 1984. Nectar extraction by hummingbirds: response to 

different floral characters. Oecologia 63: 229-236. 

, and C.L. Gass. 1981. Energy limitation of hummingbird 

populations in tropical and temperate communt ities . 
Oecologia 50: 162-163. 

, J. McA. Eadie, and L.D. Harder. 1984. What do foraging 

hummingbirds maximize? Oecologia 63: 357-363. 

, and C.A. Redsell. 1980. A nesting hummingbird feeding 



solely on arthropods. Condor 82: 463-464. 

Murcia, C. 1987. Estructura y dinamica del gremio de 
colibries (Aves : Trochilidae) en un bosque Andino . 
Humboldtia l(l):29-63. 

Murray, B.C., Jr. 1978, Interspecific territoriality in 
hummingbirds. Auk 95: 613. 

Nagy, K.A. 1980. CO2 production in animals: analysis of 
potential errors in the doubly labeled water method. 
American Journal of Physiology 238R: 466-473. 



211 



1983. The doubly labeled water (ShhI^o) method: 



guide to its use. Publication No. 12-1417, 45 pp., Univ. 
CA, Los Angeles . 

. 1987. Field metabolic rate and food requirement 

scaling in mammals and birds . Ecological Monographs 57 : 
111-128. 

. 1989. Field bioenergetics : accuracy of models and 

methods. Physiological Zoology 62: 237-252. 

, and D.P. Costa. 1980. Water flux in animals: analysis 



of potential errors in the tritiated water method. 
American Journal of Physiology 238R: 454-465. 

Neter, J., W. Wasserman, and L. Kutner . 1985. Applied linear 
statistical models. Second edition, Richard E. Irwin, Inc. 
Homewood, IL. 

Ostheim, J., and W. Rautenberg. 1989. Autonomic and 
behavioral temperature regulation as a part of the 
response complex to food scarcity in the pigeon. In: C. 
Bech and R.E. Reinertsen (Eds.) Physiology of cold 
adaptation in birds. NATO ASI Series, Vol. 173, Plenum 
Press, New York. 

Paladino, F.V. 1989., Constraints of bioenergetics on avian 
population dynamics. Physiological Zoology 62: 410-428. 

Parrish, J. 1987. Kleptoparasitism of insects by a broad- 
tailed hummingbird. Journal of Field Ornithology 59: 128- 
129. 

Paton, D.C., and F.L. Carpenter. 1984. Peripheral foraging by 
territorial rufous hummingbirds: defense by exploitation. 
Ecology 65: 1808-1819. 

Pearson, O.P. 1954. The daily energy requirements of a wild 
Anna Hummingbird. Condor 56: 317-322. 

Pennycuick, C.J. 1968. Power requiremients for horizontal 
flight in the pigeon, Columba livia. Journal of 
Experimental Biology 49: 527-555. 

Perrson, L. 1985. Asymmetrical competition: are larger 

animals competitively superior? American Naturalist 126: 
261-266. 

Petersen, R.G. 1977. Use and misuse of multiple comparison 
procedures. Agronomy Journal 69: 205-208. 



212 



Powers, D. R., and K. A. Nagy . 1988. Field metabolic rate and 
food consumption -by free-living Anna's hummingbirds 
{Calypte anna). Physiological Zoology 61: 500-506. 

Price, A.H. 1958. Vapour pressure of tritiated water. Nature 
London 181:262. 

Pyke, G.H. 1978. Optimal foraging in hummingbirds: testing 
the marginal value theorem. American Zoologist 18: 739- 
752. 

. 1981a. Why hummingbirds hover and honeyeaters perch. 



Animal Behavior 29: 861-867. 

. 1981b. Honeyeater foraging: a test of optimal foraging 

theory. Animal Behavior 29: 873-888. 

. 1981c. Optimal foraging in hummingbirds: rule of 

movement between inflorescences. Animal Behavior 28: 889- 
896. 

. 1984. Optimal foraging theory: a critical review. 

Annual Review of Ecology and Systematics 15: 523-575. 

Rashotte, M.E., and D. Henderson. 1988. Coping with rising 
food costs in a closed economy: feeding behavior and 
nocturnal hypothermia in pigeons . Journal of the 
Experimental Analysis of Behavior 50: 441-456. 

, , and D.L. Phillips. 1989. Thermal and feeding 



reactions of pigeons during food scarcity and cold. In: C. 
Bech and R.E. Reinertsen (Eds.) Physiology of cold 
adaptation in birds. NATO ASI Series, Vol. 173, Plenum 
Press, New York. 

Remsen, J.V., F.G. Stiles, and P.E. Scott. 1986. Frequency of 
arthropods in stomachs of tropical hummingbirds. Auk 103: 
436-441. 

Riechert, S.E. 1988. The energetic costs of, fighting, 
i^jnerican Zoologist 28: 877-884. 

Root, R.B. 1967. The niche exploitation pattern of the blue- 
gray gnatcatcher. Ecological Monographs 37: 317-350. 

Roskaft, E., T. Jarvi, M. Bakken, C. Bech, and R.E. 

Reinersten. 1986. The relationship between social status 
and resting metabolic rate in great tits {Parus major) and 
pied flycatchers (Ficedula hypoleuca) . Animal Behavior 34: 
838-842. 

Schmidt-Marloh, D., and K.-L. Schuchmann. 1980. Zur biologie 
des Blauen Veilchenohr-Kolibris {Colibri coruscans) . Bonn. 
Zool. Beitr. Heft 1/2: 61-77. 



213 



Schmidt-Nielsen, K. 1975. Animal physiology: adaptation and 
environment, Cambridge University Press, London. 

, and B. Schmidt-Nielsen. 1952. Water metabolism of 



desert mammals. Physiological Review 32: 135-166. 

Schoeller, D.A., C.A. Leitch, and C. Brown. 1986. Doubly 

labeled water method: in vivo oxygen and hydrogen isotope 
fractionation. 7\merican Journal of Physiology 251R: 1137- 
1143. 

Schoener, T.C. 1983. Field experiments on interspecific 
competition. American Naturalist 122: 240-285. 

. 1986. Mechanistic approaches to community ecology: a 



new reductionism? American Zoologist 26: 81-105. 

Schuchmann, K.-L. 1979. Energieumsatz in Abhangigkeit von der 
Umgebungstemperatur beim Kolibri Ocreatus u. underwood! 
(Trans: Energetic responses to various ambient 
temperatures in the hummingbird Ocreatus u. underwood!) . 
Journal fur Ornithologie 120: 311-315. 

, and H. Jakob. 1981. Energy expenditure of an 



incubating tropical hummingbird under laboratory 
conditions. LeGerfaut 71: 227-233. 

, and D. Schmidt-Marloh . 1979a. Metabolic and thermal 

responses to heat and cold in streamertail hummingbirds 
(TrochHus polytmus and TrochHus sdtulus) . Biotropica 
11: 123-126. 

, . 1979b. Temperature regulation in non-torpid 



hummingbirds. Ibis 121: 354-356. 
, , and H. Bell. 1979. Energetische Untersuchungen 



bei einer tropischen Kolibriart (Amaz!l!a tzacatl) . 
Journal fur Ornithologie 120: 78-85. 

Siegel, Sidney. 1956. Nonparametric statistics for the 

behavioral sciences. McGraw-Hill Book Co., New York, xvii 
- 312 pp. 

Smith, A. 1926. Fruit-eating hummingbirds. Condor 28: 243. 

Snow, B.K., and D.W. Snow. 1972. Feeding niches of 

hummingbirds in a Trinidad valley. Journal of Animal 
Ecology 41: 471-485. 

Snow, D.W., and B.K. Snow. 1980. Relationships between 
hummingbirds and flowers in the Andes of Colombia. 
Bulletin of the British Museum (Natural History) 38: 101- 
139. 



214 



Sokal, R.R., and F.J. Rohlf. 1981. Biometry. Second edition, 
W.H. Freeman and Co., New York, xviii - 859 pp. 

Southwick, E.E., and A.K. Southwick. 1980. Energetics of 
feeding on tree sap by ruby-throated hummingbirds in 
Michigan. American Midland Naturalist 104: 328-333. 

Stiles, F.G. 1971. Time, energy, and territoriality of the ' 

Anna hummingbird {Calypta anna) . Science 173: 818-821. j 

. 1975. Ecology, flowering phenology, and hummingbird f 



pollination of some Costa Rican Heliconia species. Ecology 
56: 285-301. 

. 1978. Temporal organization of flowering among the 

hummingbird foodplants of a tropical wet forest. 
Biotropica 10: 194-210. 

. 1980. The annual cycle in a tropical wet forest 

hummingbird community. Ibis 122: 322-343. 

. 1981. Geographical aspects of bird-flower coevolution, 

with particular reference to Central America. Annals of 
the Missouri Botanical Garden 68: 323-351. 

. 1985. Seasonal patterns and coevolution in the 



hummingbird-flower community of a Costa Rican subtropical 
forest. In: P. A. Buckley, M.S. Foster, E.S. Morton, R.S. 
Ridgely, and F.G. Buckley (Eds.) Neotropical ornithology: 
757-787. American Ornithologists' Union, Ornithological 
Monographs No. 36. 

, and L.L. Wolf. 1970. Hummingbird territoriality at a 

tropical flowering tree. Auk 87: 467-491. 

, and . 1973. Techniques for color-marking 



hummingbirds. Condor 75: 244-245. 

Tatner, P., and D.M. Bryant. 1986. Flight cost of a small 
passerine measured using doubly labeled water: 
implications for energetic studies. Auk 193: 169-180. 

Tilman, D. 1987. The importance of the mechanisms of 

interspecific competition. American Naturalist 129: 769- 
774. 

Tooze, Z. J., and C. L. Gass . 1985. Responses of rufous 

hummingbirds to midday fasts . Canadian Journal of Zoology 
63: 2249-2253. 

Waser, P.M. 1981. Sociality or territorial defense? The 
influence of resource renewal. Behavior, Ecology and 
Sociobiology 8: 231-237. 



215 



Weast, R.C. 1970. CRC handbook of chemistry and physics. The 
Chemical Rubber Co., Cleveland. OH. 

Weathers, W.W., and F.G. Stiles. 1989. Energetics and water 
balance in free-living tropical hummingbirds. Condor 91: 
324-331. 

Webster, M.D., and W.W. Weathers. 1989. Validation of single- 
sample doubly labeled water method. American Journal of 
Physiology 256R: 572-576. 

Wilkinson, L. 1987. Systat : the system for statistics. 
Systat, Inc., Evanston, IL. 

Wolf, L.L. 1970. The impact of seasonal flowering on the 
biology of some tropical hummingbirds. Condor 72: 1-14. 

. 1978. Aggressive social organization in nectarivorous 



birds. American Zoologist 18: 765-778. 

, and F.R. Hainsworth. 1971. Time and energy budgets of 

territorial hummingbirds. Ecology 52: 980-988. 

, , and F.B. Gill. 1975. Foraging efficiencies and 



time budgets in nectar feeding birds. Ecology 56: 117-128 
, , and F.G. Stiles. 1972. Energetics of foraging: 



rate and efficiency of nectar extraction by hummingbirds. 
Science 176: 1351-1352. 

, F.G. Stiles, and F.R. Hainsworth. 1975. Ecological 



organization of a tropical, highland hummingbird 
community. Journal of Animal Ecology 45: 349-379. 

Zar, J.H. 1984. Biostatistical analysis. Prentice-Hall, Inc., 
Englewood cliffs, NJ, xiv - 718 pp. 



BIOGRAPHICAL SKETCH 

Harry Morgan Tiebout III was born to Lee W. and Harry M. 
Tiebout, Jr., on November 20, 1951. An avid naturalist since 
his youth, Harry was active in the Audubon Society and the 
Isaac Walton League, as well as Boy Scouts and the Explorers. 
He remains perplexed at the long hours he has spent indoors 
watching animals in cages. Harry received his B.A. in 
psychology in 1972 at the University of Illinois. While an 
undergraduate he took two courses in ethology which forever 
changed his destiny. In 1978 he returned to school at the 
University of Wisconsin at Madison to complete a m.ajor in 
zoology, specializing in ecology and ethology. In 1982 Harry 
began his graduate program at the University of Florida. His 
master's bypass project involved radio-tracking snakes for 
two field seasons in Wisconsin. His interests in temperate 
ectotherms was thwarted by a totally premeditated switch to 
tropical hummingbirds for his dissertation. Naive but 
enthusiastic, Harry departed for Costa Rica on 1 March 1986. 
Somewhat wiser but infested, he returned to Gainesville 22 
July 1988. Harry's interests, though appearing diverse, are 
nearly all excuses to engage in his favorite pursuit, which 
shall remain lidless, legless, and nameless. Traveling, 
hiking, canoeing, camping, nature photography, and bird- 
watching are the most oft-stated alibis. 

216 



I certify that I have read this study and that in my 
opinion it conforms to acceptable standards of scholarly 
presentation and is fully adequate, in scope and quality, as 
a dissertation for the degree of Doctor of Philos-ophy. 




B^ter Feinsmger, C, 
Professor of Zoology 



ir 



1^ 



I certify that I have read this study and that in my 
opinion it conforms to acceptable standards of scholarly 
presentation and is fully adequate, in scope and quality, as 
a dissertation for the degree of Doctor of Philosophy. 



L- 



'7 / 7c'~^<-^<^ic 



// 



-"g--c-c ---i^. - 



Martha L . Crump ^ 
Professor of Zoology 



I certify that I have read this study and that in my 
opinion it conforms to acceptable standards of scholarly 
presentation and is fully adequate, in scope and quality, as 
a dissertation for the degree of Doctor of Philosophy. 




^^-g^ . 



-i5«Jg^ \F . Eisenberg 
Katharine Ordway Prof ess(5£^f 
Ecosystem Conservation 




I- certify that I have read this study and that in my 
opinion it conforms to acceptable standards of scholarly 
presentation and is fully adequate, in scope and quality, as 
a dissertation for the degree of Doctor of Philosophy. 




hU^ 



.llywhite 
Professor of" Zoology 




I certify that I have read this study and that in my 
opinion it conforms to acceptable standards of scholarly 
presentation and is fully adequate, in scope and quality, as 
a dissertation for the degree of Doctor of Philosophy. 




djdhn J. Ewel 
^-•••^rof essor of Botany 



This dissertation was submitted to the Graduate Faculty 

of the Department of Zoology in the College of Liberal Arts 

and Sciences and to the Graduate School and was accepted as 

partial fulfillment of the requirements for the degree of 
Doctor of Philosophy. 



May, 1990 



Dean, Graduate School